Reddit-driven model selection engine using sentiment analysis and semantic search to recommend the best LLM for any task.
Rebuilt 3 Top-500 equity indices (VW, EW, PW) from CRSP data. Custom VW index achieves 0.996 correlation with SPY
Policy Gradient vs. DDQN agents trained via self-play. DDQN wins round-robin tournament with 58.2% win rate.
Logistic regression model predicting Twitter influence using network centrality. Includes financial lift analysis.
Scraped 400+ books to build a reader–book network with bipartite recommender system and community detection.
Deep forest green with a horizontal bar chart showing platform allocations, ROI curve, and stat chips for LP return, MIP return, and peak month.
