Embedded Files
ML, AI, Robotics
1: Deep Learning Series
Content: Interested in learning how ChatGPT and DALLE work? Join us for weekly presentations/projects covering exciting areas like neural networks, CNNs, transformers, and diffusion models! We will cover fundamentals and applications, giving you the skills to tackle real-world deep learning challenges in vision and language. Expect to spend 1 hour per week attending meetings, with extra time if you want to complete the projects. Co-hosted by MLDS
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: Basic programming experience
Group Type: Hybrid (Presentations with Optional Projects)
Level: Beginner
Mentors: Justin Sasek, Harshal Bharatia
Curriculum: https://docs.google.com/document/d/1OqHcwKAy1mQsHLsry7msEOdfLFSv4Ze3pADTOVWJQIU/edit?usp=sharing
Intro to Machine Learning and Neural Networks (3 weeks)
Intro to Computer Vision (CNNs) (2 weeks)
Intro to Transformers (3 weeks)
Miscellaneous - Diffusion, VAE, RL, etc. (3 weeks)
2: Technical Foundations of ML Research
Content: This reading group gives you the practical skills to go from understanding machine learning concepts to writing real code. You'll learn how training works under the hood, how to use PyTorch, why we need GPUs, and how to fix things when they break. By the end, you'll be ready to contribute to ML-based research projects. This group is part of the Learning Directed Operating System (LDOS) research expedition, and motivated students can get connected to research opportunities through the program.
Required Prerequisites: Any programming experience
Short Answer Required: No
Recommended Prerequisites: Basic understanding of python
Group Type: Project
Level: Beginner
Mentor: Aryan Khatri
Curriculum: https://docs.google.com/document/d/1uTgLaYxfreaWXAg3tgUKFvNUVzGcBbC7byv2YIes7XQ/edit?usp=sharing
3: Introduction to Planning and Reasoning with LLMs
Content: Learn about different strategies to plan and reason with LLMs. This cutting-edge research field is critical in both today’s robotics and agentic research, as well as many industry tasks. It is a beginner group so no experience required but basic prompt engineering experience recommended.
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: Basic prompt engineering experience
Group Type: Reading
Level: Beginner
Mentor: Harshal Bharatia
4: AI Alignment
Content: The transition from human to machine intelligence will be incredibly transformative, but it carries a risk of catastrophe as bad as human extinction. Companies like OpenAI, Anthropic, and Deepmind have all recognized this risk and have teams working on mitigating it. As the saying goes, monkeys should be careful about inventing humans. In this group, we'll be reading and implementing the latest research on AI don't-kill-everyone-ism such as AI control, weak-to-strong generalization, and interpretability.
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: Some familiarity with deep learning and transformers
Group Type: Reading
Level: Beginner
Mentor: John Dunbar
Curriculum: https://docs.google.com/document/d/1wQj2xDvso93I6cewotPDx0gxoMf8Izy8PlIi6ycq94E/edit?usp=sharing
5: Deep Reinforcement Learning
Content: In this group, we explore modern Reinforcement Learning methods as they apply to problems in robotics, games, and much more. We build on a theoretical framework of RL (either from your previous knowledge or from doing the Intro to RL DiRP in the past), and build from early architectures like DQN to modern methods in representation learning, Offline RL, and more. We discuss the complexities that arise from shifting deep learning to RL, where the data has a fundamentally different structure and breaks many existing assumptions in deep learning.
Required Prerequisites: Knowledge of the RL framework and tabular methods, along with some familiarity with deep learning
Short Answer Required: No
Recommended Prerequisites: None
Group Type: Hybrid
Level: Advanced
Mentors: Sarthak Dayal, Abhinav Peri
6: Data-Driven Robotics
Content: What lessons can be drawn from the success of LLMs, and how can these be applied to train general-purpose humanoid robots? This group will explore the challenges of scaling robotics data through teleoperation, simulation, human videos, and more.
Required Prerequisites: Basic understanding of neural networks
Short Answer Required: No
Recommended Prerequisites: Basic understanding of CNNs, Transformers, and RL
Group Type: Reading
Level: Advanced
Mentor: Justin Sasek
Curriculum: https://docs.google.com/document/d/1EtI7KccQkvTAC7UOfNyNBj0dmN03B1VHPACD6a8vgUM/edit?usp=sharing
7: Robotics and Natural Language Processing (RoboNLP)
Content: Robotics has been primarily linked with computer vision in the last few decades, and machine learning in the last decade. In the last 5 years, there's growing interest and potential to link robotics to natural language as well, to facilitate smoother interactions between humans and helper robots. We'll do a quick intro to NLP, robotics, RL, deep learning, NLP, and CV, and end with a few papers in the area. The goal is to get you interested in perhaps doing research in the area.
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: Calculus, Basic Probability
Group Type: Reading
Level: Beginner
Mentor: Albert Yu
8: Social Navigation
Content: While autonomous driving has made massive strides in structured traffic, navigating social spaces shared with humans remains a complex challenge that is actively being investigated. This group dives into the topic of social navigation, where we will read weekly papers on navigation planning in crowded/restrictive environments, as well as related topics like vision for navigation.
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: Helpful to have a high level understanding of neural networks, transformers, diffusion, etc. for better understanding of model architecture.
Group Type: Reading
Level: Advanced
Mentor: Michael Zheng
9: LLM Training
Content: The LLM Training group will be a hands-on reading + implementation group where we explore the full extent of the training pipeline of modern production LLMs (and deployment if time permits.) Each week, we will read a seminal paper in the field, and have hands on practice by implementing it into a codebase from scatch, with the end goal of creating a small end-to-end system inspired by Andre Karpathy's nanochat. This fits people who can write python comfortably and are prepared to read fairly technical papers in depth. Prior experience with PyTorch is recommended but not strictly required. One can expect to commit, on average, at least 3 hours of work a week.
Topics: datasets, architecture, optimizers, pretraining, post training/rl, evals, etc.
Required Prerequisites: PyTorch experience
Short Answer Required: No
Recommended Prerequisites: Ideally a bit of basic knowledge about transformers
Group Type: Project
Level: Advanced
Mentor: Anthony Wang
Curriculum: https://docs.google.com/document/d/1tC2j6jvbri-z0VXaV7wleZ-j3yM0V5bZU-Hcp7xXtt4/edit?usp=drivesdk
10: Systems and Machine Learning
Content: This group is about understanding how ML systems actually work beyond just training models, including OS, scheduling, memory, latency, and real-world constraints. We’ll read papers that connect ML to operating systems and hardware, and then reinforce those ideas with small projects or experiments. If you’ve taken OS or ML and want to see how they come together in real systems, this group is a good fit.
Required Prerequisites: At least one of the following:
An operating systems course
A machine learning course
Meaningful project experience in systems
Meaningful project experience in machine learning
Short Answer Required: Yes
Recommended Prerequisites: Some background in either operating systems or machine learning; Comfort with basic programming (Python or C/C++); Willingness to read and discuss technical papers
Group Type: Either - a mix of reading and small project work
Level: Advanced
Mentor: Mahin Naveen
Curriculum: https://drive.google.com/file/d/1D-ul8rQQp8mAQecQvPO5tF1VV5rx7Cb3/view?usp=sharing
Systems, Security, Architecture
11: Introduction to Operating Systems
Content: This group will cover the core components and responsibilities of an operating system through discussions over papers which illustrate novel or important designs in this area. While discussion and readings will be a central component of this group, smaller exercises may be offered to allow members to experiment with the concepts.
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: Being enrolled in or having completed a computer architecture course is helpful but I will offer relevant background information when needed.
Group Type: Reading
Level: Beginner
Mentor: Slava Andrianov
12: Introduction to Security
Content: Exploring the world of computer systems security. We will be taking a look at the history of security, including attack mechanisms, mitigations, and defenses. I plan to cover topics such as memory safety, control flow integrity, JavaScript/JIT/browsers, software fault isolation/sandboxing, and hardware security. Time permitting, we may explore other topics (open to suggestions) and/or current security news.
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: CS429 and CS439 or equivalent
Group Type: Reading
Level: Beginner
Mentor: Caleb Eden
Curriculum: https://docs.google.com/document/d/1dMWfRxFQYDTJ1Wa8avVaBybLkg98G0a5DB1uK_CnVxU/edit?usp=sharing
13: Advanced Data Structures
Content: We will read papers about new data structures and algorithms. These papers will discuss memory layouts, threading models, designing for high concurrency, cache efficiency, and more. This group is great if you love learning about performance.
Required Prerequisites: Data Structures (CS 314)
Short Answer Required: No
Recommended Prerequisites: CS 439, 331 Algorithms not required.
Group Type: Reading
Level: Advanced
Mentor: Nicolas Garza
Programming Language, Formal Verification, Theory
14: Programming Language Design and Implementation
Content: Have you ever wondered how your favorite programming language works under the hood? Join us in creating your own language from scratch! Believe it or not, if you know how to program, you already have most of the skills you need, and in the process you'll gain a greater appreciation for the abstractions that programming languages provide.
Required Prerequisites: Any programming experience
Short Answer Required: No
Recommended Prerequisites: Concurrent or past computer architecture class or equivalent knowledge.
Group Type: Project
Level: Beginner
Mentor: Michael Jennings
Curriculum: https://docs.google.com/document/d/1kJ3pRkG09CjxMLT6HdIAJhS5lhH2X1ieSUV4xgbWNqs/edit?usp=sharing
15: Program Verification
Content: We will solve self-contained problems (think Leetcode) using programs in a special verification-oriented language called Lean 4. Then, we will prove the correctness of such programs. If time permits, we will also discuss the theory behind how Lean4 works, as well as alternative theorem-proving languages such as Isabelle and Dafny.
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: Solid programming and reasoning background (discrete mathematics and data structures)
Group Type: Project
Level: Beginner
Mentor: John Jennings
Curriculum: https://docs.google.com/document/d/12uYU7Gmv1lAK_Q76KofKJLzSnjR49LCufmWhbA5DN8U
16: Decidability and its Applications in Formal Verification
Content: SMT-based formal verification tools such as Dafny are usually not very scalable and tend to be brittle. Part of the reason is that the SMT problems they generate can involve arbitrary nestings of quantifiers intermixed with theories such as arrays and arithmetic, and such problems are in general undecidable. In the past decade, people have developed various methods to restrict verification into decidable fragments of logic and have successfully applied them in various areas of verification — a promising avenue of attack against the brittleness problem that trades off some amounts of automation for human ingenuity in encoding the verification problems. This reading group focuses on learning about (and experimenting with) recent advances in decidable verification in different first-order logic fragments. We will start with the EPR fragment and understand the fragment employed by the Ivy verification tool. If we have time, we will also survey other, more obscure fragments and discuss their usage in verification.
Required Prerequisites: Basic knowledge about automata theory, first-order logic, AND formal verification
Short Answer Required: Yes
Recommended Prerequisites: Good knowledge of formal verification and have played around with SAT/SMT solvers, program verifiers, etc.
Group Type: Reading
Level: Advanced
Mentor: Ruijie Fang
17: Metric Embedding Algorithms
Content: I'll host a weekly group where we'll discuss metric embeddings. At the beginning of the semester, we'll probably start out with some readings and discussions of common techniques in the field, but as the semester goes on we can brainstorm ideas for open problems in this space and begin researching them.
Required Prerequisites: Algorithms class, solid understanding of proofs, AND solid math background
Short Answer Required: Yes
Recommended Prerequisites: They should have taken algorithms and have a solid understanding of proofs + a solid math background
Group Type: Project
Level: Advanced
Mentor: Kristin Sheridan
Quantum Computing
18: Quantum Singular Value Transform
Content: This reading group will introduce students to the foundations of quantum computing, from linear algebra and the postulates of quantum mechanics to quantum gates and circuits through the lens of the quantum singular value transform.
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: Knowledge of linear algebra and basic understanding of quantum computing (this is a nice-to-have)
Group Type: Reading
Level: Beginner
Mentor: Satvik Duddukuru
19: Quantum Computing and Bayesian Methods
Content: Curious about quantum computing? Or interested in Bayesian statistics? This group is for you! Explore both with a self-contained introduction and guided project where you will learn the basics of these fascinating fields.
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: Familiarity with linear algebra and probabilities may be of use, as well as some experience with coding - but we will introduce concepts as necessary, so any motivated participants are welcome.
Group Type: Project
Level: Beginner
Mentor: Alexandra Ramôa
Curriculum: https://docs.google.com/document/d/1sH-Mj6wxBreV9xn5-cp4p7xj86tlgon3vEMot04pVVg/edit?usp=sharing
Medicine
20: Computational Geometry in Medicine
Content: Medicine is primed for major breakthroughs in the coming years thanks to advanced computational methods. One of the major problems in the field is representing the geometry of an artery, heart, valve, etc. in an accurate manner while minimizing the geometric descriptors necessary due to limited data. This group will mainly learn how one represents the geometry and fits it to data, medical if we have time, through project based work.
Required Prerequisites: Calculus
Short Answer Required: No
Recommended Prerequisites: None
Group Type: Project
Level: Beginner
Mentor: Ben Thomas
Curriculum: https://docs.google.com/document/d/1oqf3q0rKGp2K_g-25W92JOkYF7IGcPpepn3g1wGRXjA/edit?usp=sharing
21: Cost-Effectiveness Comparison of Two Drugs Using a Simple Markov Model
Content: This project aims to allow students to combine their mathematical, reasoning, research skills into one specific health economic model. Taking two simple and well researched drugs, in this case Nilotinib and Dasatinib, and conducting research to eventually gather data and apply them in a simple Markov model. Markov models are primarily based around probability and linear algebra. While not as technical, I hope this project will allow students to apply their technical skills into a meaningful paper.
Required Prerequisites: None
Short Answer Required: No
Recommended Prerequisites: None
Group Type: Mixture of Both
Level: Beginner
Mentor: Jan Vytopil
Curriculum: https://docs.google.com/document/d/127rRQm-jH_6HllNyiR1ZnjiLEiDlHiZDaBd7Fl_AAyU/edit?usp=sharing
Miscellaneous
22: Technologies in the Market
with Hudson River Trading (HRT)
Content: Learn about technologies used in quantitative finance such as microwave transmissions and FPGAs.
Required Prerequisites: None
Short Answer Required: Yes
Recommended Prerequisites: None
Group Type: Reading
Level: Beginner
Google Sites
Report abuse