Khalifeh Basiri

Engineered a Windows diagnostics platform for Nissan and Infiniti CONSULT-I ECUs with 9600-baud serial handshakes, framed-stream decoding for up to 22 channels, guarded fault-code workflows, and SQLite/WAL session recording. Integrated OpenAI Responses with complete vehicle context, Zod-validated human-approved tools, and grounded RAG workflows for service documentation.

Published a Manifest V3 Chrome extension that captures accepted LeetCode submissions and saves source code with generated documentation to a user-selected GitHub repository. Built privacy-first GitHub authorization, duplicate prevention, resumable historical imports, and guided repository onboarding without a developer-operated backend.

Built a real-time ASL translation system that converts 21 MediaPipe hand landmarks into 63-feature prediction vectors. Served confidence-scored TensorFlow Lite inference through a Flask API with a live React webcam interface.

Built and deployed a real-time community platform with Next.js, Convex, and Clerk, implementing reactive updates, server-side authorization, threaded comments, per-user vote state, ownership checks, and cascade deletion.

Engineered bolus and automated insulin-delivery logic within a four-person C++/Qt team, including carbohydrate and correction calculations, manual and extended dosing, CGM-driven basal adjustments, safety limits, suspend/resume behavior, and alerts.