• Designed and maintained a trade simulation platform that seamlessly integrated signals from trading strategies and user-defined configurations; harnessed the Numba library to achieve remarkable code acceleration of up to 20x; executed high-speed simulations to thoroughly test and refine trading strategies, resulting in faster and more efficient analysis• Conducted strategy backtesting, optimizing parameters, and created a formula calculator for precise signal filtration; deployed strategies to AWS EC2, empowering portfolio managers for efficient strategy enhancement• Enhanced the existing C++ codebase by integrating a data subscription feature, enabling seamless access to both historical data and real-time data streams from the backend server; encapsulated into a Python package with pybind11

• Regressed on market trades data to conduct Market Imbalance Impact for Theo Price research on two proprietary Greeks using Python; achieved both statistical and economical significance for one Greek; published paper in house• Developed Pro Rata algorithm for quoting system; constructed backtest to evaluate performance on different scenarios; estimated additional PNL of 5 million by running new logic on two Treasuries products (ZN/ZF) for 4 months• Rotated on individual equity options trading desk; made/took market for TSLA, META, AMZN, AAPL, through exchanges and OTC; executed discretionary fits for various Greeks based on market information; achieved 5k daily PNL

• Developed 20+ governance (ESG) factors such as board centrality and board efficiency using corporate board data; the final factor achieved 1.79% long-short annualized return and 2.47% rank IC• Performed data cleaning, transformation, missing value imputation and exploratory analysis using Python and SQL• Evaluated the performance of each factor based on rank IC, Information Ratio, group returns and factor coverage• Performed factor winsorization, normalization and linear-regression-based factor neutralization on common factors• Proved to have low correlations between the built factor and company’s factor pool and added the factor into one of the company’s Domestic Quality models which improved the portfolio performance by 20%• Tested the robustness of proposed factor by lagging the data from 6 to 12 months and showed the consistency

• Developed a trading strategy based on earnings surprise using alternative datasets such as company consumer sentiment and credit card transactions, which achieved an annual return of 13.2% and a Sharpe Ratio of 1.41• Built a web crawler in Python to collect product reviews from Amazon, and performed data cleaning steps such as typo replacement, word tokenization, and stemming to convert raw text data into standardized bag of words• Generated sentiment score for each product using NLTK and aggregated review-level data into company-level data• Predicted companies’ quarterly sales using linear regression based on consumer sentiment and total credit card spending, and calculated earnings surprise for each company in terms of analyst forecasts• Replicated hedge funds’ investment strategies by scraping their position disclosures in 13F filings, and analyzed their holdings using common style factors such as value, momentum, size, and carry

• Managed a $45K US equity portfolio for 2 years based on factor investment approach, and achieved an average annualized return 3% higher than the S&P 500 benchmark and a Sharpe Ratio of 1.2• Built a linear factor investment model based on fundamental factors such as P/E, P/B, and market share, and technical factors such as MACD and RSI to help stock investment decisions• Implemented semi-auto event-driven trading strategies on ThinkorSwim based on a majority vote from technical indicators such as MA and BOLL, which helped avoid the deep market callback of up to 35% in 2020

Helped hundreds of campus students transit to professional careers

• Conducted lead generation in niche sectors to find target companies based on company size, revenue, and capital limit, constructed company reports and pitched to the General Partner every week• Performed financial valuation analysis on proposed companies and helped complete an $8M acquisition

• Developed a stable cryptocurrency price model which served as the theoretical basis for a published coin• Built a mining algorithm in Matlab with 20% efficiency improvement based on source code of published coins

• Assisted in CS 61B, a data structure class taught in Java.

• Implemented back-stage management system with Python to help manage comments on VeeR community with 200k DAU