PRESENTERS

Department of Computer Science & Engineering

Elements of AI in Optimization of SEM Operation

Artificial Intelligence (AI) and machine learning (ML)are becoming intensely used in various fields of electron microscopy (EM). We focus on elements of ML in optimization of standard operation procedure (SOP) in scanning electron microscopy (SEM). There are many parts of routine SEM operation that are repetitive and very time consuming. Therefore applying computational automation methods can benefit routine operations. We review flow chart diagrams for the entire SOP to identify most compelling areas for automation: test sample positioning and standard pattern recognition for optimum performance evaluation.

RESEARCHERS: Krishnanujam S., Valley Christian High School '27; Neev T., American High School '27; Sophia H., Saint Francis High School '26; Amoug K., Irvington High School '27; Jeffrey M., Milpitas High School '26; Eshan G., Milpitas High School '27

ADVISOR: Starostina, Materials Science

KEYWORDS: Optimization Of Standard Operation Procedure | Artificial Intelligence | Machine Learning | Scanning Electron Microscopy

Department of Computer Science & Engineering

Utilizing Neural Networks to Model the Structure of the Arm with Movement and Specific Tendons Towards Development in Bionic Limbs

With the increase growth of diabetes and injuries, the need for bionic limbs has been increasing and so has its technology to help patients. However, without a computational representation of the arm with specific which are otherwise over-looked, this development in prosthetics will be slow. As a result, the ability to model the arm with different degrees of movement similar to the real arm can greatly assist in the understanding of prosthetics. By understanding different datasets on limb movement, pictures for different movements, and scans of the arm, neural networks can be utilized to compile all of this information into the best possible model for innovators to use to develop and grow the applicability of prosthetics.

RESEARCHERS: Aksh P., Washington High School '26; Ratul C., Foothill High School '25; Anika Z., Mission San Jose High School '25; Vivaan W., Dublin High School '26; Sadhika P., Quarry Lane School '25; Aahna T., Dougherty Valley High School '26

ADVISOR: McMahan, Quantum Computing & Computer Science 

KEYWORDS: Engineering | Neural Networks | Computational Modeling | Prosthetics 

Department of Computer Science & Engineering
The Evaluation of Dihedral Angle Twin Boundaries in Copper

The evaluation of dihedral angles of twin boundaries enables researchers to gauge the ratio of surface energies in deferent crystallographic orientations in FCC metals. Employing 3D printing, twin boundaries in copper were simulated, facilitating deeper understanding of granular microstructure and nature of annealing twins. Characterization was done using a range of microscopy techniques- optical, electron and atomic force microscopies. Additionally, this study delves into a comparative analysis of AFM-based findings against alternative methodologies.

RESEARCHERS: Phinna Y., Washington High School '26; Saahithi S., Monta Vista High School '27; Saketh P., Dougherty Valley High School '26; Brady S., Foothill High School '25; Michael R., Sequoia High School '25; Rui Z., Irvington High School '25; Darvas G., Washington High School '25; Ayush P., Fremont High School '26; Rohan S. P., BASIS Independent Fremont Upper '25; Raghav R., Archbishop Mitty High School '26; Avish Z., American High School '26; Arjun K., Quarry Lane '25

ADVISOR: Starostina, Materials Science

KEYWORDS: Grains l Twin Boundaries l Twinning Planes l Interfacial Free Energy l Microscopy l Copper l Fcc Metals

Department of Computer Science & Engineering

An Overview on the Performance of Reasoning Agents in Large Language Models 

The recent rise of Large Language Models (LLMs), which are able to generate human-like text, has put a large amount of attention onto AI and its potential uses. However, most LLMs are limited to a one-dimensional/left-to-right method of decision-making that can impede their performance in tasks that require accurate foresight and reference to previous decisions to execute. We hypothesize that various types of LLM reasoning agents have different strengths and weaknesses that allow for applications for different strategic use cases. In our research, we hope to determine the specific use cases and strengths of various reasoning agents, which will allow for the creation of LLMs tailored towards certain tasks with the use of such agents. With the help of reasoning agents, such as symbolic, arithmetic, and chain-of-thought reasoning, LLMs adopt a greater understanding of the context given to them and use a multi-step approach to adequately solve problems. Existing challenges in evaluating reasoning agents within LLMs include issues such as dataset biases and the potential brittleness of the model. These challenges, combined with the ethical concerns surrounding the reasoning agents such as their susceptibility to amplifying biases within a response, offer a rich research area. Using a quantitative analysis of several reasoning agents within a controlled environment, we apply diverse multi-modal and iterative reasoning techniques. Through this analysis, we explore the strengths and weaknesses of these reasoning techniques, resulting in a better understanding of the reasoning capabilities to be applied to real-world scenarios and products.

RESEARCHERS: Nathan M., Valley Christian High School '25

ADVISOR: Subramaniam, Data Science

KEYWORDS: Artificial Intelligence | Large Language Models | Natural Language Processing | Reasoning Agents 

Department of Computer Science & Engineering

Levers to improve your stock investment strategies

This research focuses on optimizing the performance of stock investment using a diverse portfolio of computational and mathematical strategies. With return and risk as the key success metrics, we developed and evaluated a few methods that aims to improve our investment results. The insights from these studies and effective levers we are building will help us move towards the high return and low risk outcomes - the "treasure island" goal of our stock investment.

RESEARCHERS: Evan L., Stratford Preparatory '26; Pranav P., Lynbrook High School ‘25; Sharvil P., Cupertino High School ‘27; Gatik G., Mission San Jose High School ‘26; Arnav S., Cupertino High School ‘25; Annika S., Shivam Academy ‘25

ADVISOR: Qin Lab Data Science & Finance

KEYWORDS: Finance | Data Science | Stock Investment | Market Prediction | Sentiment Analysis

Department of Computer Science & Engineering

Using Quantum Machine Learning to Improve Detection of Spinal Injuries, Defects, and Illnesses

Incorporating image identification machine learning models in the medical field has prospects for improving diagnostic accuracy and efficiency, quality of individualized treatment, and research productivity.  However, the complexity of the subject leaves many areas that have yet to be thoroughly and specifically researched, such as the classification of magnetic resonance images of spinal cords and injury detection.  This research targets osteoarthritis, the abnormal growth, damage, and gradual degeneration of connective tissues, and its pathologies include osteophytes, foraminal stenosis, other forms of vertebrae misalignment or collapse that are likewise detectable through magnetic resonance imaging.  By employing a quantum convolutional neural network and visual transformer, we optimize classical computing solutions by taking advantage of the increased complexity, compression, and precision of quantum computing.  

RESEARCHERS: Alyssa C., George Walton Comprehensive High School '26, Mahika R., Dougherty Valley High School '26, Lucas C., Archbishop Mitty High School '25, Ryan J., Foothill High School '25, Jason N., BASIS Independent Silicon Valley Upper School '26, Nelson N., BASIS Independent Silicon Valley Upper School '26, Srijon M., Amador Valley High School '25, Avni S., Irvington High School '25

ADVISOR:  McMahan, Quantum Computing & Computer Science 

KEYWORDS:  Computer Science | Machine Learning | Quantum Computing | Convolutional Neural Networks | Health technology

Department of Chemistry, Biochemistry & Physics

Machine Learning to Unravel the Dynamics of Chaotic Systems such as the Double Pendulum

The prediction of chaotic systems is known to be very important to the field of physics. Understanding the movements and behavior of chaotic systems could potentially revolutionize various fields such as aviation. The ability to be able to predict stalls directly results from the prediction of chaotic systems would be able to predict chaotic movements of winds that affect the amount of lift under the wing. Furthermore, the control of chaotic systems that results from being able to predict these systems could result in more efficient power delivery systems and turbines as well as assist the healthcare industry with more efficient defibrillators and brain pacemakers. Artificially stimulated chaotic brain waves could someday help inhibit epileptic seizures as well. The prediction of chaotic systems could fundamentally alter how industries function.

RESEARCHERS: Vasista R., Milpitas High School '25, Aviram J., Washington High School '25, Arnav B., Dougherty Valley High School '26, Sai H. R. N., Leigh High School '26

ADVISOR: Akl, Machine Learning for Condensed Matter Physics

KEYWORDS: Machine learning | Neural Networks | LSTMs | Chaotic Systems | Physics | Classical Mechanics | Double Pendulum

Department of Computer Science & Engineering

The Development of an AI-Enhanced Autonomous Drone for Emission Prediction and Pollution Monitoring

Our research involves implementing machine learning algorithms into an autonomous drone to reduce pollution levels and greenhouse gas emissions. We plan to train our algorithm with numerical data consisting of the concentration of nitrogen, CO2, and methane molecules for atmospheric conditions. After training, the drone will collect testing data from the atmosphere and predict the concentrations of these greenhouse gasses. Based on the concentration in a given area, the drone will adjust its navigation toward locations of high concentration to determine possible emission sources. With these findings, we can implement clean energy sources to reduce emissions and better our environment. The aim is to use the drone’s predictions to determine the areas that need the most assistance in reducing emissions. The drone will continuously record the pollution levels for different areas and give highly updated readings based on the greenhouse gas concentrations. Currently, manual pollution monitoring is costly, ineffective, and inefficient. By constantly giving accurate readings with the concentrations, our autonomous drone limits the need for manual testing, which saves time and allows us to find solutions for highly polluted areas. 

RESEARCHERS: Shivesh S., Dougherty Valley High School '25, Karthik S., Mission San Jose High School ‘26, Avi K., Irvington High School ‘27, Shanay G., Dublin High School ‘26, Sahiel B., Dublin High School '26

ADVISOR: McMahan, Quantum Computing & Computer Science 

KEYWORDS: Autonomous Drone | Artificial Intelligence | Machine Learning | Pollution Monitoring | Pollution

Department of Biological, Human & Life Sciences

Designing and developing a recombinant mRNA vaccine against the Nipah virus (NiV) by targeting fusion protein binding between mammalian and viral glycoproteins

The Nipah Virus (NiV) is a single-strand antisense RNA virus that relies on the specific binding of receptor proteins to increase its transendothelial migration across mammalian cells. The binding mechanism of NiV attachment protein G with EphrinB2 and Ephrin B3 activates F-mediated fusion, resulting in infection of the cell. However, current limitations in drug discovery efforts against NiV focus on the use of Remdesivir and Ribavirin and need to take into account the virus's innate ability to mutate, consequently rendering most organic small molecule approaches ineffective as long-term prophylactic agents. Here, we show the potential usefulness of a messenger RNA (mRNA) vaccine library as a stand-alone or combinatorial therapeutic agent along with FDA-approved drugs. Using in silico and in vitro models, recombinant NiV vaccine libraries were synthesized to replicate mutated receptor attachment glycoproteins. By using a codon optimization model, we evolve the NiV attachment proteins in an evolutionary congruent fashion. These candidate mRNA vaccines were encapsulated in lipid nanoparticles (LNPs) and transfected into mammalian cells to assess the efficiency of transfection and viability of the vaccine. Our early-stage mRNA vaccine library presents great potential since it is well-substantiated that these glycoproteins are required for viral entry. Similar to SARS-CoV-2, NiV exhibits rapid community transmission, however, with a higher mortality rate of 40-75%. This study proposes the effectiveness of a highly scalable and potentially low-cost mRNA vaccine library to circumvent the consequences of a NiV outbreak.

RESEARCHERS: Rajesh V., Saint Francis High School '24, Caleb Y., Saratoga High School '25, Anishka D. Basis Independent Silicon Valley '25, Shreya R. Monta Vista High School '25, Erin L., Irvington High School '25

ADVISOR: Amadi Lab Biotechnology and Synthetic biology 

KEYWORDS: Biotechnology | Vaccine Development | Codon Optimization

Department of Biological, Human & Life Sciences

Examining Correlations Between Creative Technological Design and Consumer Cognitive Dissonance

In an era where technology and artificial intelligence (AI) are seamlessly integrated into daily life, understanding cognitive dissonance in this context is pivotal. This study investigates cognitive dissonance through a novel approach, focusing on the dichotomy between consumer belief and action in response to various creative technological product designs. Participants were presented with paired statements, each reflecting opposing values in categories such as autonomous vehicle and video game design, new smart home technologies, and the innovative use of AI in healthcare. Our study was developed to promote participant creativity as they put themselves into theoretical situations and evaluate the degree to which they are comfortable with various AI-based technologies. Ratings on a scale of 1 (strongly disagree) to 10 (strongly agree) were obtained for each statement, and a 'dissonance score' was calculated based on the absolute difference between these paired ratings. Theoretically, larger differences indicate lower dissonance, as participants demonstrate consistency between their beliefs and actions. Conversely, smaller differences suggest higher dissonance, denoting contradictions between their beliefs and actions. Our preliminary analysis indicated a potential correlation between cognitive dissonance towards the design of autonomous vehicles and video games among adult demographics. Additionally, our initial data indicate a gender-specific trend, with male participants exhibiting a higher propensity for dissonance in response to the implication of AI technologies in healthcare. These observations warrant further, more detailed analysis to reveal the underlying implications of innovative technological design. our study offers new approaches to understanding consumer cognitive dissonance in the rapidly evolving landscape of innovation-based technology and creative AI software.

RESEARCHERS: Arya S., Dougherty Valley High School '25, Moksha R., Mission San Jose High School ‘25, Avigna S., Dublin High School ‘24

ADVISOR: Jahanikia, Computational Cognitive Neuroscience, AI & Data science

KEYWORDS: Cognitive Dissonance | Behavioral Plasticity | Technology | Product Design | AI | Smart Homes | Autonomous Vehicles | ChatGPT

Department of Computer Science & Engineering

Predictive Study on Sales of Cardio Good Fitness Treadmills

Currently, cardiovascular health is becoming more crucial to maintaining good heart health. As a result, in order to maintain it, many people choose to do cardiovascular exercise by running on treadmills. Cardio Good Fitness is a retail store that sells fitness equipment like treadmills. In this study, the goal is to use machine learning to predict the preferences of this business's customers. To do this, we will be analyzing customer behavior by looking at various factors like their demographics and past purchasing behaviors to create a model that can predict the customers' preferences when buying treadmills. This will greatly help the business of Cardio Good Fitness by helping them optimize their product offerings.

RESEARCHERS: Helen L., Saint Francis High School ‘25, Aneya S., California High School ‘25, Camille C., The Nueva School ‘27

ADVISOR: Dharmale, Electrical Engineering

KEYWORDS: Treadmills | Data Analysis | Machine Learning | Demographics

Department of Biological, Human & Life Sciences

Uterine Cancer Drug Delivery

Coming soon.

RESEARCHERS: Vishnu C., Washington High School '26, Mahati S., American High School '26, Maanya S., American High School '26, and Nehal R., Dublin High School '26

ADVISOR: Amadi Lab Biotechnology and Synthetic biology 

KEYWORDS: Coming soon.

Department of Chemistry, Biochemistry & Physics

Machine Learning of an EKG dataset

Coming soon.

RESEARCHERS: Amish G. Vista Del Lago '27, Jonathan G., Saratoga High School '25, Kunal S., American High School, '25, Lucas S., Lynbrook High School '27, Mansi M. American High School '25, Nishant J., Evergreen Valley High School '25, Sophia H., Columbia High School '25

ADVISOR: Akl, Machine Learning for Condensed Matter Physics

KEYWORDS: More info to come.

Department of Biological, Human & Life Sciences

Optimizing Arthrospira Platensis Growth with Other Cyanobacteria Species for Bioenergy Production

Anthropogenic activities are the driving force behind climate change, specifically through the massive amounts of greenhouse gases that are released in this activity. As a result, bioenergy, specifically biofuel, has gained the attention of many scientists as a possible solution to alleviate humans’ environmental impact. In this context of addressing the increasing demand for sustainable solutions and investigation in biofuel, one noteworthy avenue is the utilization and cultivation of cyanobacteria. Cyanobacteria, also known as blue-green algae, have gained significant attention in both the global market and scientific research. These microorganisms offer multifaceted advantages, as they can be harnessed for the production of various goods, biofertilizers, medicines, supplements, and namely biofuel. The study investigates Arthrospira Platensis, a highly valuable genus of cyanobacteria, through co-culturing it with other cyanobacteria genera, Nostoc and Anabaena, which are both nitrogen fixing cyanobacteria species which have developed unique symbiotic relationships with other organisms in their natural habitats. Through experimentation, it is evident that Arthrospira Platensis genus benefits from being grown with other species as opposed to being cultivated by itself. 

RESEARCHERS: Gauri R., Amador Valley High '25, Aahana L., Dougherty Valley Highschool '26, Shreshta J., Dougherty Valley High School '26, Audrey H., Saint Francis High School ‘26

ADVISOR: Kaur, Microbiology & Environmental Genetics

KEYWORDS: Cyanobacteria | Symbiotic Relationships | Bioenergy | Microbiology

Department of Chemistry, Biochemistry & Physics

Discovery, Synthesis, and Optimization of 5-phenylisoxazole Based Covalent Inhibitors Targeting G12C Mutant KRAS for the Treatment of Cancer

Oncogenic mutations in the GTPase protein KRAS are implicated in approximately 25% of human cancers. Specifically, the G12C mutation, a common mutation found in KRAS-related pathology, is found in 12% of non-small cell lung cancers and 3% of colorectal and other solid tumors. This single residue substitution causes irreversible binding of GTP/GDP to the catalytic site, thereby forcing the protein into a permanent, activated state. While KRAS has been previously considered an undruggable chemotherapeutic target, the discovery of acrylate-based covalent inhibitors of G12C KRAS has led to the development of two FDA-approved drugs: Sotorasib (AMG-510) and Adagrasib (MRTX849) which inspired our own pharmacophore model, and our library of isoxazole-based covalent inhibitors of G12C KRAS. En route, we optimized a previously reported amide coupling in which our library of analogs exhibited a comparatively higher yield of 98%. This transformation tolerates air with a trivial loss of yield and has been applied to 12 different examples of arylmethyl isoxazole acids and alkyl substituted piperazines including in the synthesis of Nucleozin. In vitro potency was then evaluated through MTT assays against Calu-1 cancer cell lines, and to test the selectivity, against HCT-116 cancer cell lines. Our S-methyl compounds were shown to be selective in targeting mutant G12C, as they were ineffective in HCT-116 colon cells, and our lead compound in specificity amongst these contains a 2,6-dichloroaryl ring.Medicinal Chemistry

RESEARCHERS: Arshia D., Saratoga High School '25

ADVISOR: Njoo Synthesis | Physical Organic Chemistry | Catalysis | Chemical Biology | Spectroscopy | Medicinal Chemistry 

KEYWORDS: Medicinal Chemistry

Department of Biological, Human & Life Sciences
Humans vs LLMs: A quantitative comparison in creativity

Challenging the prevailing perception that artificial intelligence is far less creative than humans, our study explores the creativity of Large Language Models (LLMs) by focusing on their ability to impersonate individuals. Utilizing established measures such as the Alternate Uses Task (AUT), Remote Associates Task (RAT), and Torrance Tests of Creative Thinking (TTCT), we compiled an original creativity assessment that quantifies both convergent and divergent thinking. Furthermore, our test includes the self-reported personality measure of the NEO Five-Factor Inventory (NEO-FFI) to generate an extensive profile of the participant’s creativity. Moreover, in facilitating an unbiased comparison between the LLM and human participants, we provided the LLM with the participants’ demographic data, including but not limited to race, age, gender, profession, salary, and location. Such allowed for more personalized and human-like responses from the LLM, producing a reliable comparison between the two. Our preliminary data indicates that LLMs demonstrate significant creativity, outscoring humans on all three creativity assessments while offering novel, comprehensive, and relevant responses to each test. Our preliminary observations challenge previous assumptions regarding the creativity of AI and clarify its positive relation to humans. Our findings allow for further exploration into the dynamics between humans and AI as a means to bolster human creativity. 

RESEARCHERS: Manas B., John F Kennedy High School '25, Nishka V., Foothill High School '25, Samuel L. Valley Christian High School '24

ADVISOR: Jahanikia, Computational Cognitive Neuroscience, AI & Data science

KEYWORDS: Artificial Intelligence | Large Language Models | Creativity | Divergent Thinking | Convergent Thinking | Big C | Small C

Department of Computer Science & Engineering

DFT / Machine Learning group

More info to come.

RESEARCHERS: Ashmit S., Dublin High School '25, Grace A., Basis Independent McLean '26, Kevin G., Dublin High School '25, Meadow S. Lynbrook High School '25, Praneel A., Leigh High School '26, Shiven E., California High School '26, Sunay V., Saint Francis High School '26

ADVISOR: Akl, Machine Learning for Condensed Matter Physics

KEYWORDS: More info to come.

Department of Computer Science & Engineering

Using Autonomous Drones for Terrain Mapping and Route Creation and Optimization to Enhance Search and Rescue Operations

In this project, we use an autonomous drone to map out terrain following natural disasters in order to give first responders an updated, real-time view of the ground they will be moving through. By using several sensors on the drone paired with flight code, the drone can carry out quick and efficient mapping routes to give first responders a fully optimized travel route for them that they can use in order to carry out rescue operations in a much more streamlined and efficient manner.

RESEARCHERS: Kaustubh L., Amador Valley High School '25, Aryav D., Mission San Jose High School '26, Shaunak J., American High School '26, Claire L., Basis Independent Silicon Valley High School '26

ADVISOR: McMahan, Quantum Computing & Computer Science 

KEYWORDS: Autonomous Drones | Machine Learning | Computer Vision | Terrain Mapping

Department of Chemistry, Biochemistry & Physics

Progress towards an expanded SAR of C4 modified analogs of podophyllotoxin. 

The rich diversity of lignin small molecules derived from podophyllotoxin, a non-covalent tubulin inhibitor isolated from the Podophyllum family, has led to the clinical development of several FDA-approved anticancer agents, including DNA topoisomerase inhibitors etoposide and teniposide. While these compounds share the same tetracyclic core, two subtle structural changes that differentiate podophyllotoxin from its DNA topoisomerase-binding analogs—the presence of 4’ methylation on the aromatic ring and stereospecific glycosylation at the C-4 hydroxyl—yield two independent mechanisms. Given the immense pharmacological importance of these two features, we sought to establish a structure-activity relationship regarding modification at C-4 on the potency, specificity, and chemical properties of podophyllotoxin. Here, we synthesized a systematic library of 21 podophyllotoxin analogs with analogous ester, carbonate, and carbamate substitutions. The antiproliferative activity and efficacy of our analogs as tubulin inhibitors was evaluated through cell viability assays, tubulin polymerization assays, computer docking models, and cell cycle analysis. Our previous efforts with esters showed that increasing C-4 bulk decreases potency against human cancer cells but insignificantly impacts cell-free assays. From our preliminary SAR, small carbocyclic carbamates at C-4 are well tolerated in cell free tubulin polymerization experiments, but large, bulky alkyl groups are less well tolerated.

RESEARCHERS: Harriet C., Las Lomas High School '24 

ADVISOR: Njoo Synthesis | Physical Organic Chemistry | Catalysis | Chemical Biology | Spectroscopy | Medicinal Chemistry 

KEYWORDS: Natural Products | Tubulin Inhibitors | Chemical Biology

Department of Biological, Human & Life Sciences
COVIDFatigue: Characterization and Severity Assessment of Post-Acute Sequelae of SARS-CoV-2 (PASC)

Long Covid has affected millions of people across America. COVID-19 patients that tested negative months ago still face physiological and neurological effects with unknown time frames. By surveying people who tested positive for COVID-19, we are able to connect the data to demographic information to discover the longevity of COVID-19 effects. Every participant is between 18-55 years, fluent in English, and has not suffered from a serious medical condition prior or during the COVID-19 infection time. Through observation of demographic data, the majority (~97%) of participants tested positive for COVID-19 after receiving vaccination. Since the vaccination rate was high, the severity of the symptoms was inversely low. We aim to expand our scope of participants to include a larger demographic group.

RESEARCHERS: Keerthana N., Evergreen Valley High School '25; Sanya K., Valley Christian High School '25

ADVISOR: Jahanikia, Computational Cognitive Neuroscience, AI & Data science

KEYWORDS: COVID | Long COVID | Brain Fog | Cognitive vs Physiological Symptoms

Department of Computer Science & Engineering

DFT / Material Science group / Epitaxy and DFT on pervoskite crystals

More info to come.

RESEARCHERS: Ezana M., Archbishop Mitty '26; Hankyu K., Bellarmine College Prep '26; Jessica S., St. Francis High School '26; Loccini G., Heritage High School '25

ADVISOR: Akl, Machine Learning for Condensed Matter Physics

KEYWORDS: More info to come.

Department of Computer Science & Engineering

Advancing Environmental Mapping and Forest Health Assessments: Integrating Machine Learning Algorithms in Autonomous Drones

With the increasing concern for environmental conservation, there is a growing need for efficient methods of environmental mapping and forest health assessments. However, traditional methods employed by the U.S Forest Health Monitoring have faced controversy due to limited spatial resolution and integration of modern technologies. This research paper explores the application of machine learning algorithms in autonomous drones to conduct forest health assessments. Autonomous drones have the ability to collect timely, up-to-date data, which offers enhanced accuracy. This study focuses on training Deep Learning (DL) models to classify different environmental features based on aerial imagery captured by drones. To achieve accurate and efficient data collection, we will utilize Red-Green-Blue imaging and Convolutional Neural Networks (CNN) with the appropriate evaluation metrics, such as the Normalized Difference Vegetation Index (NDVI), Leaf Area Index (LAI), and foliage color, to create tree classes and identify forest health indicators. By integrating machine learning algorithms into forest health assessment, this study provides a more efficient, accurate, and up-to-date approach to monitor and evaluate the well-being of forests—supporting ongoing efforts towards environmental management and conservation. 

RESEARCHERS: Andrew D., Leland High School '26, Anika M., Valley Christian High '26, Omar A. '27, Varun N. Homestead High School '26

ADVISOR: McMahan, Quantum Computing & Computer Science 

KEYWORDS: Machine Learning | Drone | Autonomous | Sustainability

Department of Biological, Human & Life Sciences

Comparative Genomic Analysis of Colorectal Cancer Microbiome Bacteria to Discover Novel Relationships

Colorectal cancer (CRC) is uncontrolled tumor growth that starts in the rectum or colon (Park E. et al., 2022). Many factors affect the development of cancer, including daily habits, environments, and genetics. Our research focuses on analyzing the differences in pathways/enzymes between cancerous and non-cancerous associated bacteria in the gut microbiome outlined by a recent cancer microbiome review (Park E. et al., 2022). By utilizing the Bacterial and Viral Bioinformatics Resource Center (BV-BRC), we compiled our bacteria’s genetic information into genome groups and used the comparative systems service to identify target pathways and construct phylogenetic trees. After focusing on genomes, we delved deeper into the enzymes. The programming language R was used to narrow down four specific enzymes from the set of genomes: two from the pathways only in non-cancerous bacteria and two in cancerous-associated bacteria. A Multiple Sequence Alignment (MSA) run at the genome level identified the range of lowest entropy among the genes in the four enzymes - one of which had the lowest range of 30-40. We are using NCBI Blast and other bioinformatics methods to characterize/validate the four enzymes in our respective target bacteria. Our end goal is to target/screen the unique pathways and enzymes (like the enzyme with EC number 5.4.3.2) of the cancer-associated bacteria and non-cancerous associated bacteria to decrease the metastasis of CRC tumors (Park E. et al., 2022). These genes, that help create the enzymes, can be manipulated in the wet lab as shown by the cited paper.

RESEARCHERS: Harshita K., Amador Valley High School '25, Anish J., Dougherty High School '24, Cheryl C., Dougherty High School '24

ADVISOR: Cunha Lab Bioinformatics and Cancer Biology

KEYWORDS: Bioinformatics | Colorectal Cancer | Gut Microbiome | Cancer Biology

Department of Computer Science & Engineering

Optimization of Error Correction on the Surface Code using Graph Neural Networks under a Bosonic Bath

The Surface Code provides robust protection from the noise and errors which are common in modern quantum systems through topological properties and has proven to be an effective framework for encoding qubits against errors introduced by the environment. However, decoding information from the syndromes measured on any Surface Code is a computationally intensive task. Currently, the most popular algorithm for decoding the Surface Code is the Minimum Weight Perfect Matching (MWPM) algorithm, which computes the shortest route of an error propagation. However, the accuracy of MWPM, and other explicit decoding models, depends code properties such as the quality of models; and estimates of error rates for idling qubits, gates, measurements, and resets; and assumes symmetric error channels (Lange, Moritz, et al.). Here, we introduce a GNN-based neural network that closely matches the performance of the MWPM algorithm, extending the application of data-driven error decoding methods for correlated noise models. The goal of this study is to demonstrate the capability of GNNs over the MWPM algorithm under data from surface codes under a bosonic bath. 

RESEARCHERS: Rohin V., Mission San Jose High School '26, Diya V., Valley Christian High School '26, Pranav G., Monta Vista High School '26

ADVISOR: McMahan, Quantum Computing & Computer Science 

KEYWORDS: Quantum Error Correction | Surface Code | GNN | Bosonic Bath | Correlated Noise

Department of Computer Science & Engineering

Theoretical study of quantum dots in noise-resistant quantum key distribution systems

Quantum key distribution (QKD) systems hold the potential for secure communication by leveraging the laws of quantum mechanics. However, challenges arising from noise and environmental interactions can compromise the security and efficiency of QKD systems. This research aims to explore the theoretical aspects of utilizing quantum dots within noise-resistant QKD systems. By leveraging advanced simulation techniques on a high-performance computing cluster, we will investigate the impact of quantum dots on enhancing the security and reliability of QKD systems.

RESEARCHERS: Prahlad S., Washington High School '25, Praneel S., Academy of the Canyons '25, Samuel L., Saint Francis High School '26, Sumedha K., Stanford Online High School '26, Xiangtuo C., Basis Independent Silicon Valley '26

ADVISOR: McMahan, Quantum Computing & Computer Science 

KEYWORDS: Quantum Dots | Quantum Key Distribution Systems | Quantum Communications

Department of Biological, Human & Life Sciences

Development of Lipopetides as a Combinatorial Treatment for Colorectal Tumors

Colorectal cancer is a common and often fatal disease that arises in the colon or rectum. It typically originates from benign polyps, which can gradually transform into cancerous cells if given enough time to grow. Some of the symptoms that may arise are constipation, blood in stool, and diarrhea; however, if these symptoms are not intercepted, the cancer cells will proliferate throughout the body and eventually lead to death. Current therapeutic measures include preemptively removing polyps before they develop into cancer cells and using chemotherapy or immunotherapy. However, these treatments are often invasive or applicable to only earlier stages of cancer.

Our research focuses on exploring the potential of lipopeptides, or a series of amino acids connected to lipids, as a novel therapeutic approach for colorectal cancer. They can be a less invasive and more effective treatment for more advanced stages of colorectal cancer. We hope to develop and study the potential anticancer properties of lipopeptides and apply them to HCT-116 (Human Colorectal Tumor) cells. Our project involves cultivating an HCT-116 cell line along with designing and developing variations of lipopeptides through printing peptides and linking them to lipids.

RESEARCHERS: Sarah Z., Irvington High School '27, Tarishi P., Mission San Jose High School '26, Aryan T., BASIS Independent Silicon Valley '26, Barghav B., Irvington High School '25, Aanya A., The Harker School '27

ADVISOR: Amadi Lab Biotechnology and Synthetic biology 

KEYWORDS: Lipopeptide Therapeutics | Colorectal Cancer | HCT-116 Cells | Anticancer | Drug Discovery | Medical Chemistry | Chemical Biology

Department of Chemistry, Biochemistry & Physics

Benchtop 19F NMR spectroscopy enables mechanistic analysis and catalytic optimization for the preparation of celecoxib and mavacoxib, 3-(trifluoromethyl) pyrazolyl benzenesulfonamides non-steroidal anti-inflammatory drugs (NSAIDs)

Organic fluorinated compounds have generated a significant precedent within medicinal chemistry, boasting metabolic stability and potent therapeutic capabilities. Of these, eighteen FDA-approved organofluorines have been classified as non-steroidal anti-inflammatory drugs (NSAIDs). NSAIDs competitively inhibit the cyclooxygenase enzyme, mediating the conversion of arachidonic acid to inflammatory prostaglandins. A quantitative kinetic analysis for fluorinated COX-2 inhibitors celecoxib and mavacoxib benzenesulfonamides has yet to be established. Real-time benchtop 19F NMR spectroscopy is used to determine absolute kinetics of catalyzed pyrazole cyclo-condensation, as well as the identification of otherwise unobservable transient intermediates. With this workflow in hand, we screened Lewis and Brønsted acids in order to enable catalytic preparation of celecoxib and mavacoxib. Ultimately, we envision future applications of this workflow in the synthesis of other fluorinated scaffolds.

RESEARCHERS: Allen K., Henry M Gunn High School '24

ADVISOR: Njoo Synthesis | Physical Organic Chemistry | Catalysis | Chemical Biology | Spectroscopy | Medicinal Chemistry 

KEYWORDS: Organic Synthesis | Methodology | Medicinal Chemistry | 19F NMR Spectroscopy

Department of Computer Science & Engineering

Diagnosing different types of skin cancer using Machine Learning (CNNs)

Skin cancer poses a significant health challenge, with early diagnosis crucial for effective treatment. Inspection by dermatologists remains the standard for diagnosis, but it can be subjective and prone to error. This project explores the potential of convolutional neural networks (CNNs) as a powerful tool for skin lesion classification, aiming to improve accuracy and accessibility of diagnosis and ultimately contribute to better patient outcomes. Leveraging the data of over 10,000 images of pigmented skin lesions categorized into seven diagnostic classes (including benign and malignant tumors), we will use a deep learning model based on CNNs. This project showcases the power of deep learning and CNNs in revolutionizing skin cancer diagnosis.

RESEARCHERS: Aryan G., West High School '25; Aarush N., Mountain House High School '25

ADVISOR: McMahan, Quantum Computing & Computer Science 

KEYWORDS: Skin Cancer | Machine Learning | Convolutional Neural Networks | Disease Diagnosis

Department of Computer Science & Engineering

Generating Chemically Stable Molecules via Quantum Computing and Python Molecular Benchmarking Processes

Current drug discovery and development processes cost billions of dollars and between five to twelve years to create one FDA-approved drug, so researchers have been implementing computational chemistry methods for current molecular synthesis pathways. However, these  methods often have inefficient runtimes, due to the large size of the chemical space. Aiming for a more efficient runtime and robust analysis of high-dimensional molecular data, our group previously implemented the Hybrid Quantum-Classical Generative Adversarial Network (QGAN), and the most recent iteration of the model is the Hybrid Quantum-Classical Graph Generative Adversarial Network (QNetGAN) to synthesize chemically feasible molecules. The QNetGAN addressed the issues faced by the QGAN, including the distance between atoms exceeding bonding length, by generating molecular graphs and utilizing long-short term memory cells. QNetGAN was able to generate 141/300 structurally valid molecules that satisfy Lipinski’s Rule of Five, yielding a 47% success rate — a notable increase from the prior 2.3% — with a minimal training time of 10.164 minutes. However, molecules generated by QNetGAN remain a work in progress, with numerous molecules failing to satisfy the Octet Rule and having unoptimized bond lengths/angles. Our current work focuses on implementing chemical post-processing algorithms to increase the chemical structures’ stability and feasibility. Our Octet Rule algorithm employs a Depth-First Search traversal algorithm to count the number of bonds and open orbitals on each atom. Meanwhile, our formal charge calculation algorithm calculates the formal charge of each atom to find the most stable structure for the molecule . Furthermore, our Hydrogen Addition algorithm builds upon the Octet Rule algorithm to traverse through the molecule’s adjacency matrix, adding hydrogens to central atoms where necessary to complete their outer shells. Most recently, we are implementing algorithms to convert our generated molecular graphs from adjacency matrices to .XYZ files which provide 3D coordinates for each atom in a molecule, such that the final molecular structure corresponds correctly to their molecular geometries and bond angles. We also created an algorithm to detect cyclic compounds, which will be handled differently by our molecular geometry algorithm. Although a work in progress, our post-processing algorithms have been able to check for Octet Rule satisfaction, calculate formal charge, and add hydrogen atoms to molecules as necessary with 100% accuracy. In the future, we plan to continue testing our algorithms on larger molecular structures, implementing methods to handle exceptions to the basic rules, and integrating the optimization algorithms into the QNetGAN training process which we have begun to explore through another model, QNetGAN 3.0. Furthermore, we will continue to explore other types of models, such as RNN and VAE.

RESEARCHERS: Adelina C., Archbishop Mitty High School '24, Max C., Sir Winston Churchill Secondary School '24, Linda C., Homestead High School '25, Hasset M.,Piedmont Hills High School '24, Leena A., Irvington High School '27, Falak C., Mission San Jose High School '24, Aditya P., The Quarry Lane School '27, Diya J., The Quarry Lane School '25

ADVISOR: McMahan Lab Quantum Computing & Computer Science 

KEYWORDS: Drug Discovery | Quantum Computing | Machine Learning | Generative Adversarial Network

Department of Computer Science & Engineering

Project Treasure Island - Finding Strategies to Build the Best Stock Investments

This research focuses on optimizing the performance of stock investment using a diverse portfolio of computational and mathematical strategies. After defining the key performance metrics of stock investments: return vs. risk, we analyzed the relationship between these two metrics across 30+ investment strategies designed by the student researchers. We compared the investment outcome with our researcher's intuition on risk estimation when they designed the strategies. Towards improving the investment performance, we studied the effects of diversifying the portfolio and frequent strategy updates through experimentation and analysis. Moving forward, we will design prediction models and trading algorithms to drive our investment results towards high return and low risk outcomes - the "treasure island" goal of this project.

RESEARCHERS: Eliana H., Dougherty Valley High School '25, Brian L., Quarry Lane High School '25, Annika S.,Shivam Academy '25, Sharvil P.,  Cupertino High School '27

ADVISOR: Qin Lab Data Science & Finance

KEYWORDS:  Stocks | Finance | Investment | Data Science | Risk vs. Return | FinTech

Department of Biological, Human & Life Sciences

Molecular Characterization of Vernonia Amygdalina

Vernonia amygdalina (V. amygdalina) commonly known as bitter leaf plant, is a small West Africa shrub belonging to the daisy family that has been used in a variety of traditional African medicines to treat malaria, fevers, gastrointestinal issues, diabetes, amongst other ailments. Despite its extreme bitter taste chimpanzees have been observed eating V. amygdalina when they are experiencing gastrointestinal issues. We explore the biomolecular properties to determine whether our plant of interest is purely anecdotal or is beneficial to human health. There have been a number of phytochemical studies evaluating V. amygdalina for clinical applications. However, there is a lack of publicly available data detailing the identification of small organic molecules, proteins, and ribonucleic acids (RNA) that would be useful in identifying apoptogenic compounds, if any, that would support evidence of clinical or nutritional benefit. Here we show that V. amygdalina contains a variety of anti-cancer, anti-inflammatory, anti-diabetic, and antimalarial drugs, such as Azetidine, Undecane, and Lapachol. Using GCMS, we show that this plant contains a number of small organic compounds FDA-approved that have a variety of medicinal properties and few with unknown benefits. By making our biomolecular database of V. Amygdalina publicly accessible, future researchers of this plant can use this library to aid in their research. Moreover, this library will also aid in this project's next steps of evaluating select compounds against various cancer cell lines to determine if there are unknown additional benefits.

RESEARCHERS: Arjun S., The Quarry Lane School '25, Medha R., Irvington High School '27, Krishna D., Franklin High School '27, Desiree P., Homestead High School '24

ADVISOR: Amadi Lab Biotechnology and Synthetic biology 

KEYWORDS: Biomolecules | Medicinal Chemistry | Africa | Biology | Bitter Leaf | Anticancer | Antimalarial

Department of Computer Science & Engineering

Python-based Hartree-Fock approximations for molecular energy

The ground state energy of a molecule is very useful in verifying the feasibility and stability of a given molecule. This energy is difficult to compute exactly, but approximation methods can be used to obtain very close estimates of the energy. One such method, called Hartree-Fock, is simulated with a Python program called PySCF to calculate these energies. Our research aims to verify the accuracy and efficiency of this program and make it more accessible for use in the future through the creation of a UI to streamline the research process.

RESEARCHERS: Allen L., Monta Vista High School '25, Ram S., Liberal Arts and Science Academy '25, Akash G., McMahan Lab '26

ADVISOR: McMahan Lab Quantum Computing & Computer Science 

KEYWORDS: Quantum chemistry | molecular energy | Hartree-Fock | PySCF

Department of Biological, Human & Life Sciences

Development of Synthetic Aptamers for Use as Low-cost PLD1 Inhibitors

PLD1 is a gene in the genome of the cell that codes for an enzyme that breaks down phosphatidylcholine into phosphatidic acid and choline. The phosphatidic acid then leaves the cell and attaches to an mTOR receptor in another cell and causes a signal to be sent into the cell. As a result of the signal transduction and a phosphorylation cascade, the cell starts performing mitosis. Due to excessive amounts of phosphatidic acid the mTOR receptor constantly signals to the cell to start mitosis through DAG kinases. PLD1 amplifies anti-apoptotic functions that cause inflammation and create a roadblock for chemotherapy. The focus of our project is to create synthetic aptamers using specialized DNA or RNA bases to act as competitive inhibitors for PLD1 allowing patients to continue chemotherapy.

RESEARCHERS: Anushka S., Mission San Jose High School '24', Kavya D., Washington High School '26', Mansha G., Washington High School '26', Divya G., Dougherty Valley High School '26', Izna K., Saratoga High School '26'

ADVISOR: Amadi Lab Biotechnology and Synthetic biology 

KEYWORDS: Synthetic Biology, Cancer Biology, Biotechnology

Department of Chemistry, Biochemistry & Physics  

Evaluation of Bio-inspired Ionizable Lipids for Lipid Nanoparticle mRNA Delivery

More info to come

RESEARCHERS: Kimberly K., Amador Valley '24

ADVISOR: Njoo Lab Synthesis | Physical Organic Chemistry | Catalysis | Chemical Biology | Spectroscopy | Medicinal Chemistry 

KEYWORDS: 

Department of Biological, Human & Life Sciences

Development of Lipopeptides as Bioactives for Drug-Resistant Fungi

Fungi are a diverse group of microorganisms that play critical roles in various ecosystems. Certain fungal diseases have a significant impact on agriculture, leading to losses in crop yields and food shortages. Appearing in the early 20th century and in the 1990s, a new strain of Foc (Fusarium oxysporum f. sp. cubense), known as Tropical Race 4 (TR4) or Panama 4, emerged in Southeast Asia and has since spread to other regions, including Africa and Central America and pushed a number of crop variants toward extinction. Lipopeptides are a class of natural and synthetic compounds consisting of a peptide chain linked to a lipid moiety. They exhibit a wide range of biological activities, including antimicrobial, antifungal, anticancer, and immunomodulatory properties. Recent advances in fungal research have provided insights into the mechanisms of fungal infections and potential targets for antifungal therapies or biopesticides. However, more research is needed to develop effective treatments and prevent the emergence of drug or chemical-resistant fungal strains. Our project involves the design and development of lipid and peptide molecules; including the modeling of gene product targets and synthetic molecules, designing and printing peptides, and lipid linkage.

RESEARCHERS: Netra T., Monta Vista High School '25', Nishika D., Washington High School '25', Maya C., Basis Independent Fremont Upper School '27', Srinidhi V., Irvington High School '27'

ADVISOR: Amadi Lab Biotechnology and Synthetic biology 

KEYWORDS: Chemical Biology | Drug Discovery and Development | Biotechnology | TR4 | Fungal Disease | Lipopeptide Therapeutics

Department of Computer Science & Engineering

Using Quantum Neural Networks (QNNs), Quantum Vision Transformers (QVT), and the Mathematical Morphological Reconstruction Algorithm (MMR) for Brain Tumor Detection

Brain tumors affect millions around the world, so detection is critical to helping doctors determine treatment. Currently, radiologists manually identify tumors through MRI (Magnetic Resonance Imaging) scans; however, this poses several limitations: it creates a heavy reliance on the experience of radiologists, has become increasingly costly and time-consuming, and is not as accessible to areas that lack the necessary resources and doctors. With the advancement of deep learning algorithms, a more accessible and efficient solution is possible. Given the existing research in classical Convolutional Neural Networks (CNNs) for tumor detection, Quantum Convolutional Neural Networks (QCNNs) and Quantum Vision Transformers (QVT) offer a promising approach to the problem. Mathematical Morphological Reconstruction (MMR), another image processing method, provides a relative metric for success in the QCNN, and is another classical alternative to CNNs. This research compares the accuracy and computational speed of the MMR, QCNN, QVT, and CNN algorithms to determine whether introducing a quantum aspect presents any noticeable advantage. To build these models, extensive datasets of MRI brain scans were collected. The MMR algorithm involved applying various techniques such as dilation, erosion, and skull stripping through OpenCV2's morphology functions. The QCNN algorithm utilizes quantum power to encode the data into a parametrized quantum circuit and apply convolutional and pooling layers. In terms of future steps, QVTs will be implemented with QCNNs for higher spatial understanding. So far, our results indicate that the MMR algorithm achieved up to 92% accuracy. These results will be compared with the accuracy of the QCNN, QVT, and CNN algorithms.

RESEARCHERS: Tiffany Liu Quarry Lane '25, Riddhi Sharma Evergreen Valley '26, Eesha Gadekarla Quarry Lane '26, Shivansh Bansal Dublin High '25, Ryan Li  Basis Independent Fremont '25

ADVISOR: McMahan, Quantum Computing & Computer Science 

KEYWORDS: Quantum Computing | Brain Tumors | Machine Learning| Mathematical Morphological Reconstruction | Convolutional Neural Networks