A self-evolving dual-LLM framework where one model injects faults and another learns to localize them, generating 5,000+ synthetic C++ programs to train a 12B-parameter debugger.
Founding engineering work at Lab Systems: an Electron + Python disk forensics application that delivered 70% faster evidence processing through distributed C++ kernels and a unified investigator UI.
Cryptocurrency forensics platform built at Lab Systems for tracing illicit on-chain activity. Distributed RPC ingestion, ML-based clustering heuristics, and a sleek investigator UI shipped to forensic clients.
Real-time environmental visualizations driven by sensor data and Stable Diffusion. Achieved a 90% success rate aiding decision-making in user studies.
Emotion-driven adaptive UI for e-learning that drove a 30% increase in user engagement by tailoring interface state to detected affect.
Blockchain-based pharmaceutical supply chain platform for tamper-proof drug traceability. Improved traceability by 80% and reduced counterfeit instances in pilot deployments — granted German patent.
AI-driven mask compliance and automated attendance system using YOLO-based detection on distributed camera streams.
