On NETRA-AI, I formulated counter-drone engagement as a real-time multi-effector deep reinforcement learning problem, bringing together RF jamming, kinetic intercept, and cyber-takeover. I stress-tested the model across 500+ synthetic scenarios for the iDEX Open Challenge.
For RakshaChain, I built a digital-forensics chain-of-custody pipeline on Hyperledger Fabric, with cryptographically enforced role-based access and client-side verification using the WebCrypto API. The live prototype was deployed for the MeitY / C-DAC Blockchain India Challenge 2026.
I’ve also built agentic systems including ReflexionOS, where I fixed a degenerative feedback loop and added a live reasoning-trace visualizer, and ComplyAct, which combines document extraction, UI navigation, and a rule-based escalation to human review.
As a DevOps Intern at Elevate Labs, I automated CI/CD pipelines and introduced canary deployments, reducing manual deployment effort by 20%. I also built Kubernetes-isolated sandbox environments and monitoring dashboards that reduced incident response time by 30%.
