Eddie Fu

Go-to-Market Team

Bay Area Founders Club (BFC), founded in Silicon Valley in 2022 and managed by a dedicated group of Stanford alumni and students, is a thriving community of over 50,000 members, including more than 2,000 startups and over 650 venture capital firms. With most startups in our network raising $1-10 million, and some exceeding $20 million, our mission is to empower creators who aspire to make a global impact. We provide the essential resources, support, and connections needed to transform visionary ideas into reality. Follow us to stay updated!

Go-to-Market Team

TA for CIS 5500: Database & Information Systems Structured information is the lifeblood of commerce, government, and science today. This course provides an introduction to the broad field of information management systems, covering a range of topics relating to structured data, from data modeling to logical foundations and popular languages, to system implementations.

• Developed a transformer-based multi-encoder anti-money laundering (AML) model using NVIDIA Merlin and Hugging Face LLMs. Engineered the model to process 180-day histories of card transactions, payments, and bank transactions for comprehensive risk analysis. • Achieved robust model performance with AUC of 0.9371, recall of 0.9722, and false positive rate of 0.1673 on the out-of-time test dataset. Demonstrated model robustness on new customer data, with AUC of 0.8691, recall of 0.8511, and false positive rate of 0.2992. • Developed targeted investigation strategies based on comprehensive risk profiling, resulting in a 6.82x increase in precision while maintaining a recall of 0.7, and at the same time identifying potential new AML risk sequences among below-the-line transactions.

• Used Python to scrape marketing information, sales data, and product reviews from Douyin and Dianping, integrating them with the CRM system. Automated daily processes on cloud servers and achieved a 95% reduction in data collection time. • Created databases in MySQL and HDFS, built interactive dashboards with Flask to show key performance indicators. Built XGBoost, LightGBM and Random Forest models to find key features. Improved online shop exposure by more than 25%. • Harnessed OpenAI's gpt-4 and whisper models to automate categorization and compliance checks of Apple's text, audio, and image data on enterprise WeChat. Resulted in 30% cost savings for Apple China's online customer service channel.

• Designed dashboards in Tableau to provide visual insights into key operational metrics. Automated weekly reports generation and streamlined weekly payment systems using Python, leading to a 15% reduction in payment processing time. • Optimized ML models to identify high CLV customers and influential creators on Apache Spark and Hive. Improved memory, reduced shuffling, leveraged Hive partitioning and addressed data skew, boosting cost efficiency by over 50%. • Leveraged Neo4j graph visualization to detect and remove authors manipulating view counts. This innovative strategy elevated genuine video views by over 30% and concurrently achieving a significant 30% reduction in associated costs.

• Conducted an in-depth financing analysis for enterprises in the medical sector. This involved web scraping historical financing data for thousands of companies and subsequently visualizing the data to provide insight for M&A decisions. • Created databases of potential target companies in MySQL on cloud servers, built models using Python to predict the relationship between corporate financing and company survival, and completed screening of potential target companies. • Played a pivotal role in acquiring two companies—a prominent gynecology hospital group and an online psychiatry medical company. Executed due diligence, crafted investment proposals, and presented to the investment committee.

• Distilled insights from extensive reports to formulate trading strategies. Explored multi-factor stock selection approaches, harnessing data related to fundamentals, forecasts, and text and image contents to identify more than ten influential factors. • Deployed a diverse set of machine learning methodologies—ranging from tree-based, boosting and stacking ML models to deep learning models like RNN, LSTM, and advanced NLP techniques—for constructing stock return predictive models. • Engineered a suite of portfolio constructing models, including risk budgeting and Moving Average timing. Refined factor selection through backtesting the portfolio's Sharpe ratio, ultimately assembling a portfolio of Sharpe ratio exceeding three.

Summer Program—Designing Innovation to Action: Life, Leadership, and AI Program

Transferred to Penn Engineering after finishing the curriculum.