Multi-stage image provenance pipeline combining face detection, ArcFace representations, reverse-image discovery, metadata extraction, and deterministic evidence fingerprinting. Integrated Google Lens discovery through SerpApi and Ethereum Sepolia anchoring to produce independently verifiable provenance records with explicit failure and blocked states.

• FastAPI
• InsightFace
• SerpApi
• Ethereum Sepolia

Offline-first LLM benchmarking platform for evaluating local models across quality, performance, and resource constraints through standardized workloads. Built around Ollama with structured validation, reproducible benchmark artifacts, hardware profiling, and model-selection workflows — no cloud LLM dependencies.

• Python
• Ollama
• LLM Benchmarking

Full-stack F1 race intelligence platform combining Monte Carlo simulation, historical race analysis, live telemetry, ML models, and retrieval-augmented intelligence for strategy decisions. Developed an AI race-engineer workflow with explainable strategy recommendations, contextual memory, telemetry analysis, and simulation-backed pit-wall decision support.

• React
• FastAPI
• RAG
• XGBoost
• Docker