The computer vision library holds 19 measured projects on its own page. Next to it: an order-execution engine, an AI SaaS, a production file-conversion API, an Android app and sports analytics.
From classical OpenCV to fine-tuned detectors, segmentation, 3D and foundation models. A CVPR paper reimplemented from scratch, a detector deployed 17× faster on an NPU, and the debugging stories behind each number.
Slices large orders over time (TWAP, VWAP, POV, Almgren-Chriss), routes each slice across Kraken and Coinbase by all-in price, and checks every fill against hard risk limits in C++. Deterministic clock for exact replays. Paper trading only.
Tailors a resume and cover letter to each job, then grades it against the ATS with the missing keywords. Security-first architecture: OAuth 2.1 (PKCE), token-derived resource paths (IDOR-safe), default-deny rules, signature-verified billing webhooks. Also a remote MCP server for AI agents.
Production REST API for batch conversion of 7+ formats (PDF, DOCX, HEIC, TIFF), up to 1 GB per upload; Factory pattern for strategy selection.
Role-based Android booking app for 3 user types with conflict prevention and notifications, tested with JUnit on CircleCI.
Does how fast a player builds speed in the first second of the 40-yard dash predict how he moves in NFL games, beyond the official 40 time? A reproducible tracking pipeline, cohort-held-out models and an honest answer: mostly no.
Feature-engineering pipelines (Pandas, scikit-learn) feeding a filterable React risk dashboard over real financial data.