Courses

Personal feelings towards some of the recent courses that I took and would recommend, by no means accurate descriptions of the material.

Spring 2023

• CS 144: Introduction to Computer Networking. Great system class that dissemble the Internet into pieces.
• CS 155: Computer and Network Security. (Disclaimer: only done on a VM.)
• STATS 315B: Modern Applied Statistics: Learning II. A refresher on ML methods.
• MATH 237A: Topics in Financial Math: Market microstructure and trading algorithms. Final project in development on pairs trading with ML.

Winter 2023

• CS 228: Probabilistic Graphical Models: Principles and Techniques. A deep dive into learning and infering distributions using PGMs that completely brushes my mind on Bayseian learning methods.
• CS 238: Decision Making under Uncertainty. Taking the class while learning basic Go from my friends.
• MATH 238: Mathematical Finance. Pricing models built on SDE.

Fall 2022

• CS 149: Parallel Computing. After the tortures of learning to parallelize almost everything, one would become more skeptical and itchy about every line of code that runs happily on a single thread.
• STATS 305A: Applied Statistics. Statistical testing and procedures taught in a formal way with, surprisingly harder, more than 1 dimension. Many cool observations are distributed throughout the problem sets and coding projects.
• Math 215A: Algebraic Topology. Fundamental groups, homologies, and cohomologies of real and imagined shapes.

My Younger Self

• AI intro classes: CS224N, CS231N, CS246. In these classes with big names, I got to implement a question-answering model and an image aesthetics scoring model. Good (but not enough) take on NLP and CV. CS246 (Mining Massive Data Sets) was a great class to dip into a compilation of data mining methods, from recommending systems to page ranks to GNNs.
• CS Theory classes: CS154 (Introduction to the Theory of Computation), CS166 (Data Structures).
• Math classes to chew: Math 122 (Modules and Group Representations), Math 154 (Algebraic Number Theory), Math 205A (Real analysis).

Work

AutoDL — A Python based, open-source Auto Deep Learning (AutoDL) API as a backbone architecture and starter-tool for a series of AutoDL competitions, aiming to pipeline the process of running AutoDL experiments. Repo links (developing): AutoDL Competition Submission, AutoDL Workflow.

Chart Question Answering Workshop — Worked on quarter-based machine learning challenges and workshops in Stanford ACM Club. Collaborated with other club members at the CVPR 2021 Chart Question Answering Challenge and used CNN, OpenCV and pytesseract OCR tool to identify bar charts and pie charts.