A Graduate Student in Computer Science

Even though it was unimaginable for me, as a child, to become a proud member of the Computer Science community one day, it happened.

I believe that Machine Learning (ML), much like an undergraduate student, can be applied to a wide range of fields, even those that it initially has no knowledge of. So I have learned ML tools and techniques to apply them to different subjects, innovate new ideas, and solve various problems.

Beyond my major, I like to work in various fields including Art, Business Administration, Cognitive Science, Economics, Mathematics, Medical Care, Neuroscience, Physics, Psychology, and Sociology. Likewise, I am eager and hope to combine all these sciences and disciplines with ML, if possible.

As I am not a first-generation student, education is always one of the highest values in my farsighted family, as I am usually encouraged to study to guarantee my future against any crisis. My mother always says, "The only thing that can never be taken away from us is what we carry in our minds." Although it was hard for me as a boy to choose studying over watching Spider-Man or Batman animated series, my intrinsic ability to understand mathematics, other sciences, and skills motivated me to learn and go further. This perspective encouraged me to engage in many fields such as sports, learning how to play chess and swim, music, where I learned to read notes, play the xylophone and then the flute, and English as my foreign language. I studied English for more than five years at the Iran Language Institute (ILI), previously known as Iran America Society, which is the most prestigious, oldest, and largest foreign language teaching institute in Iran. I graduated from this institute with a high-intermediate certification of completion.

Instead of childish dalliance, I studied and entered NODET. Despite having problems that tainted my childhood, I still believe I did the right thing. I remember begging my Geography teacher to add 0.25 to my score so I could be considered among the top students. Unfortunately, he refused. After that experience, I decided to change my mindset regarding my study goals. While enhancing my learning ability, my kind grandfather suddenly died. After that, tenderhearted me, as a family kid, stopped studying and educating well until the next academic year; death in my family (comfort zone) was new to me! I renewed myself and started working hard at learning but with weak motivation: Just for self-assertion. Although I succeeded in that way, I, as a deliberate but patient teenager, decided to change my core beliefs and motivations. One year later, I began self-examination and my psychological journey, thanks to my mother, who is a psychotherapist. I started learning guitar, which helped me discover my musical interests and significantly trained my ears. This led me to participate in music events at NODET, both playing guitar and singing! I also have a passion for literature, which I combined with my love for music and rhythm, gaining extensive experience in versification. I was awarded first place in Mazandaran Province Poetry Festival.

I gradually realized that it was better to deepen my knowledge in Mathematics and Physics, so I began preparing in this field for the Iran University Entrance Exam (Konkour-Concours). Later, I was admitted to the University of Tehran, the best university in Iran according to the latest U.S. News & World Report magazine ranking. Worth saying, it was challenging for me as a teenager to completely control myself and follow my noteworthy values, like academic ones, instead of adolescent distraction. After all, I did it as well.

Although my main educational interest was mathematics, I also considered industry, technology, and business, leading me to study Computer Science, because it was the intersection between Pure Mathematics and Computer Engineering. (If not an engineering program) This is why I did not go for Computer Engineering, which is the intersection of Computer Science and Electrical Engineering. I thought that the process of entering a well-known university, where we can learn as much as possible and utilize such a university, should be the last hardship in my life. !Notwithstanding, it was just the beginning

The methods and approaches of learning in schools are indeed vastly different from those in universities: During my school years and the first semester of university, I would wait to be taught or prompted to learn. I attended classes and tried to answer any questions asked, but I often completed my homework without much discipline, and sometimes, I did not do it at all. I was suddenly taught much about algorithms, computer architectures and programming, and the Russell Paradox in Set Theory, which were hard for me to understand and strengthen myself as well. Despite all that has been mentioned, the tragic disaster of "Ukraine International Airlines Flight 752" caused immense dismay among everyone, including myself. As a result, I struggled to find the energy to focus on learning and preparing for the final exams. Anyway, you had better not search the Middle East on Google.

Besides the lacks of caution, I found out that something needed to change in me. I sought new methods of gaining knowledge with my fellow students and found several approaches, such as self-studying "Calculus 1 & 2" and "Data Structures." Just when we were making strides—knock, knock—the COVID-19 pandemic hit the door! Online educating, a great number of daily fatalities, distrust of some teachers, and so on. These were the main challenges that added to my prior concern over education at university.

I passed "Linear Algebra" with almost 20 out of 20 completely by myself to protect my truthfulness while other students did their best to cheat. This approach generally helped many students to enhance their Grade Point Average (GPA) drastically once and for all. Some rules had become more usual those days: During passing the "Fundamentals of Combinatorics" (online!) course, while I had promised myself not to cheat, the chief TA announced that students who had cheated would receive their score, and he gave me zero for writing nothing!

Above all, I tried to adapt myself to the new situation. The adaptation might be slow, but it happened. I did trial and error to learn as well. I gradually became more and more successful in passing my courses, usually with acceptable grades of A or B. I tried not to enroll in many main or high-level courses as they would not be taught in such a great way. Therefore, the only prominent courses that I learned about and passed during those two years were "Advanced (Python) Programming," "Design and Analysis of Algorithms," "Fundamentals of Combinatorics," "Graph Theory and Applications," "Linear Algebra," and "Probability 1." I found Python appealing due to its simplicity and practicality, and have since started using it.

I continued to participate in psychological courses and workshops, exploring Transactional Analysis and Schema Therapy. I participated in Finance, thanks to my father, who is a stockbroker. I then tried trading cryptocurrencies, but unfortunately, I failed and lost all my money. This experience taught me that entering any field requires prior education and experience. After that, I achieved success in recouping the lost money and even making profits beyond the initial deposits.

Not only in Finance, but I also began exploring areas where I could excel, such as Music, Cinema, Sports, and Social Health. In music, I hired a voice coach and continued this collaboration after discovering my talents, adding basics of Music Theory and Solfège. I also advanced to intermediate guitar learning. In Sports, I took up tennis and bodybuilding to achieve a fitter body and stronger muscles. Additionally, I trained with Electronic Muscle Stimulation (EMS) for one year to enhance my muscle resiliency, body functioning, and overall strength. In Social Health, I joined the Red Crescent community for online courses to educate myself about pandemic diseases and human health. These few were the only benefits of the dark pandemic period, and I have planned to continue in these fields after completing the application for this university. This is why I am multifaceted and have a rich résumé despite some inefficiencies.

Anyway, after two tough years, I finally attended university classes again! Everything was good except the night before the "Introduction to Theory of Computation" final exam. I became sick and had to drop the course to prevent my GPA from including a 0 score, despite having an acceptable background in the course. I handled this special situation independently and will never forget it. The next semester came, and the first week was great and brought something new for me. I began my first teaching assistant position. However, in the second week, after innocent Mahsa Amini was killed just because of some ideological conflicts, many blatant protests happened on a large scale, especially at University of Tehran, where I was spending most of my study time. Again, I was forced to change my way of education, but with the prevailing frightening atmosphere, as many of my fellow students were damaged in the protests or out of them, I struggled to perform my best, particularly in my favorite subjects, "Artificial Intelligence" and "Strategic Games I," which unfortunately impacted my overall GPA. Facing so much vicissitude in those days and years, I used to answer the "How is it going?" question with this statement: "I just have not become a father yet."  During those dark days, I started being psychoanalyzed to embark on a journey of self-discovery and to shape a clearer, more informed worldview, more than any time before.

After the people's exasperation and frustration became dormant and Deep Learning models became more usable and applied than in the past, like Large Language Models (LLMs) like ChatGPT, I decided to delve deeper into Machine Learning and its applications. My motivation for education came back to me. Then, I enrolled in Non-Linear Programming, which was a recollection of those days in which I completely enjoyed studying matrix calculus and calculations, and convex optimization problems. Additionally, I learned about some algorithms like Gradient Descent, which is the base of learning algorithms for neural networks. I experienced a lot and became exhausted. However, I continued my education specifically about ML, as I believe it will create a new species of living organisms and I enjoy having a part in such a field.

I extended my university education to five years to deepen my knowledge of ML and do anything I could have done for my education process before. I enrolled in three ML-related courses, namely "Advanced Information Retrieval," "Bio-Computing," and "Data Mining." Passing all such courses with outstanding grades verified my eagerness and understanding of ML and Deep Learning. I used to be a teaching assistant more than usual to learn how to teach students the subjects in the simplest comprehensible way and express myself in class until my graduation. At the end of graduation, I was unofficially regarded as the best Teaching Assistant ever in the department students' minds. Professor Sadri, whose head teaching assistant I was most, kindly certifies this belief, too. I enrolled in and officially completed many ML-related Coursera and LinkedIn Learning online courses. Learning from international fecund professors and instructors is always interesting to me. I participated in writing a paper to compare various drug synergy score prediction methods that helped me develop critical thinking, effective search techniques, and scientific writing skills and learned how to read articles, summarize their main points, and understand the structure of writing articles. This helped me develop critical thinking, effective search techniques, and scientific writing skills. Under the supervision of Professor Sajedi, I wrote my undergraduate thesis on the application of BERT in my native language. It was exciting to see ML applications in my local context. After my unofficial graduation, I wrote a paper on the efficiency of Bidirectional Recurrent Neural Networks. This experience inspired me and taught me a lot about research and its primary challenges, such as finding related works and evaluating and concluding new results that had not been stated before. Regarding to these research experiences, Professor Sajedi is probably proud of me, as we have planned to collaborate on new topics in the future. I joined the first artificial intelligence laboratory called 'AAILAB' at the department of the university as a research assistant, under the direction of Professor Nadi. This has been my considerable teamwork and studying between many areas compactly called "Financial ML." I did many simple and complicated tasks and sought new frameworks and financial or ML techniques, too. After all, I joined the Persian Music Laboratory (PeM Lab), directed by Professor BabaAli, where I participated in labeling a massive music dataset. The fact that both laboratory directors are pleased with my performance and I did not fail, withdraw from, or pass with an unsatisfactory grade any of the courses related to my field of further work in my Ph.D. studies, especially those related to ML, makes me proud.

Currently, I am exploring opportunities for admission into competitive master’s or Ph.D. programs. My research interests lie at the intersection of ML and interdisciplinary applications. Building on this perspective, I am especially eager to pursue ideas that integrate ML with the following fields:

✔️Psychology, Neuroscience, and Social Network Analysis: I am particularly interested in the intersection of ML and Psychology. One line of inquiry I am exploring is the development of LLM-based systems that can serve as specialized therapists or psychoanalysts, trained on foundational theories such as Freud’s structural model, Jung’s archetypes and shadow, MBTI, ACT, schema therapy, transactional analysis, and related frameworks. Rather than merely suggesting solutions, such a model would aim to understand, express, and respond to human emotions authentically—remaining transparent and consciously aware rather than masking or simulating feelings.

From a social-science perspective, I am also interested in building models that gently challenge and broaden users’ beliefs and perspectives instead of reinforcing echo chambers, while accounting for cultural differences and ethical implications. Within this scope, I am curious to analyze the factors that drive the popularity of certain public figures—for instance, Aisan Eslami, a Persian blogger who has amassed over 15 million followers, the largest verified audience among Iranians. As a self-study effort, I have also begun learning the basics of neuroscience.

Since both Psychology and Neuroscience study how the brain shapes thought, decision-making, and behavior, I believe that combining insights from both fields and applying them to domains such as Social Network Analysis or Music Information Retrieval could yield excellent results. To capture this vision, I have coined the term “Neuropsyence”—and in an optimistic sense, I aspire to grow into a “Neuropsyentist.” In the short term, my goal is to investigate these directions more deeply, combining rigorous empirical methods with ethical design to create intelligent systems that foster insight, growth, and well-being.

✔️Music Information Retrieval, Neuropsyence, Recommendation Systems, Music Theory, and Singing: Having listened extensively to vocal, instrumental, and soundtrack music, I have developed a distinct classification approach for each type. Vocal and soundtrack pieces can be categorized based on the emotions they convey, guided by a model that differs from classical emotion frameworks such as Russell’s (1980) and Thayer’s (1991). Should I transfer my idea and map it into these popular emotion classification models? No one knows yet! For instrumental pieces, classification may depend on the singer historically associated with the composition.

As a freelance singer, I believe that with some talent and musical awareness, we can reproduce almost all notes and chords within the range of our larynx—our uniquely gifted high-level instrument. A model that considers larynx health could assist in developing this range safely. Likewise, many official and unofficial references exist for the chords of songs we wish to sing or play. Some closely match the original track, some follow music theory rules but differ from the original, and some are incorrect. A large model trained on songs whose chords have been officially published by composers and arrangers could greatly assist singers and musicians in accessing accurate original chords. We could also develop specialized chord-extractor models for each instrument, considering instrument-specific musical modeling.

As a social and artistic application, I would also like to train a model capable of diagnosing natural singing talent. Candidates would perform without the use of autotune, background effects, or backing tracks, allowing the model to evaluate their raw vocal tone, control, and expressiveness. Such a system could support music educators, producers, and enthusiasts in identifying genuine vocal potential. 

Building on these intuitions, I am developing a multilingual framework that classifies music based on emotions and sentiments extracted from lyrics, metadata, and audio signals—using mBERT for lyric semantics and CNN/RNN architectures for audio modeling. Beyond classification, I am exploring the use of neuro-wave signals to dynamically match listeners with music whose frequencies and waveforms align most closely with their current state. As a highly imaginative idea in Neuroscience and brain science applications for music, I dream of creating a system that allows users to capture their desired memories, creative thoughts, or dreams and transform them into music videos—for personal, artistic, or blogging purposes.

Furthermore, I aim to develop a specialized Large Language Model that can engage in conversations about emotions, music theory, music history, genres, singers, composers, legendary figures in music, etc. This model would also recommend music aligned with users’ preferences—for example, suggesting songs similar in genre or mood to the ones they already love.

Taken together, these ideas point toward a comprehensive and personalized music recommendation system that integrates psychological, neuroscientific, and musical considerations to enhance emotional resonance and practical usability in music experiences.

✔️Computer Vision, Recommendation Systems, Natural Language Processing, Photogrammetry, and E-Commerce: I am also interested in applying ML to next-generation e-commerce experiences. I am envisioning a user-centered shopping platform powered by computer vision and language models. Early ideas include 3D avatars for virtual clothing try-ons, face analysis for personalized makeup suggestions, and an LLM assistant that can chat about diet, preferences, and recommend products. While still exploratory, I am excited to investigate how these elements could work together to create a more intuitive, personalized, and engaging shopping experience. I aim to expand this pipeline so that any product category can be matched and tried virtually using multimodal signals, and am mindful of the technical and ethical challenges—data requirements, fairness, privacy, and robustness—and believe these must be central to system design. In the short term, I plan to prototype core modules such as the 3D try-on and conversational assistant, evaluating them on realism, recommendation relevance, and privacy-preserving performance.

✔️Economics and Finance: I am also interested in the application of ML to Economics and Finance. To develop a comprehensive model capable of predicting prices, I believe it is essential to integrate four complementary facets: microeconomics and macroeconomics, fundamental analysis, technical analysis, and sentiment analysis (with special attention to social media). My idea is to construct this system in a structural way—first by designing submodels specialized in each facet, and then combining their outputs through a weighted integration mechanism. While I recognize that building and training each submodel would demand significant financial, computational, and human resources, my short-term goal is to investigate and prototype elements of this framework. In doing so, I aim to advance toward more robust, transparent, and multifaceted predictive models for complex financial systems.

✔️Healthcare and Medical Science: I am deeply motivated to apply ML to Healthcare, focusing on diseases with high mortality rates such as cancer and Alzheimer’s. I am aware of significant progress in tumor detection, cell-level malignancy prediction, and early neurodegeneration diagnosis, yet the persistence of these diseases highlights the need for more effective solutions. My short-term goal is to contribute to research that improves early detection and diagnosis through learning models capable of identifying subtle biomarkers and risk indicators. By combining technical rigor with interdisciplinary collaboration, I aim to advance the application of ML toward reducing mortality and improving patient outcomes.

✔️Fitness and Nutrition Science: After history, Fitness and Nutrition Science may be one of the most dynamic and uncertain fields of study: one day, experts recommend running or doing aerobic exercise before breakfast, and the next, they advise eating beforehand. Beyond this constant evolution, each individual’s body follows its own unique principles when it comes to metabolism, adaptation, and health response. Motivated by this variability, I am developing the idea of a complex model that records users’ physical changes, daily routines, preferences, and goals, while learning their dietary habits and exercise behaviors over time. By combining this information, the model could generate adaptive and highly personalized fitness and nutrition plans—plans that evolve dynamically as the user’s body and habits change. Such a system would integrate multimodal data including physiological metrics, lifestyle signals, and psychological patterns to provide recommendations that go beyond generic health advice. In essence, I aim to design an intelligent companion that learns each person’s biological and behavioral rhythm, helping them reach their fitness goals safely and effectively through data-driven, individualized insight.

As a self-assigned academic activity, I have started self-studying neuroscience through scientific free videos and exploring reinforcement learning using Reinforcement Learning: An Introduction (Second Edition) by Richard S. Sutton and Andrew G. Barto. Above all, I roam through Coursera and LinkedIn Learning in search of new and applicable courses—to review whatever I have already learned or known, and of course, to discover something new and delve deeper, even if only a few centimeters, into the vast sea of knowledge surrounding my major, ML, its applications, and other sciences that may inspire it. At the same time, I am exploring my passion for teaching by recording lessons on subjects I know well and can explain with clarity and enthusiasm. Likewise, I sing songs that resonate deeply with me, pouring my emotions into each performance. The results of both pursuits find their home on my separate YouTube channels, where I share them with the world.

In summary, I aim to address both theoretical and real-world challenges across fields where I find inspiration, have innovative ideas, or am invited to contribute. In this regard, I have been developing several directions that I hope to explore further during my graduate studies.