Published open-source security software and a trading engine built for live deployment. I'd rather have one tool people actually run than ten polished demos.
RedCell actively exploits MCP servers and maps what it finds to the OWASP MCP Top 10. Understanding the offensive side is what makes the defensive side real.
Python backends, React frontends, infrastructure like Redis and worker queues, and ML research down to quantum circuits. I go where the problem is.
I've run technical sessions for 250+ members and judged hackathons. Communicating hard ideas simply is part of the work, not an afterthought.
MCP security scanner that actually exploits
An open-source scanner that proves vulnerabilities in Model Context Protocol servers by calling their tools with real attack payloads — path traversal, SSRF, OS command injection, secret leakage. Findings map to the OWASP MCP Top 10. Ships as a CLI, REST API, and web dashboard with HTML/SARIF reports and a GitHub Action.
Multi-strategy NSE algo trader
An algorithmic trading engine for the Indian market (NSE) with momentum, moving-average, and volatility strategies. Full backtesting, risk management, and a live paper-deploy path via the Dhan broker API — built toward a real go-live.
Serverless functions with a kinetic UI
A serverless Python function-execution platform with an auto-scaling worker pool, Redis-backed job queue, and a real-time telemetry dashboard. Workers spawn dynamically under load; the kinetic monolith UI visualizes jobs and events as they happen.
Quantum circuits meet VAE + GAN
Research combining variational autoencoders and GANs with quantum circuits, showing a 3.7× reconstruction improvement over classical baselines on MNIST. Includes an ablation study identifying 6 qubits + circular entanglement + 2–3 layers as the optimal configuration.
I want to build security and systems work that quietly gets adopted — open-source tools people reach for, and infrastructure that just works at scale. Nearer term, that means growing RedCell into a real adaptive red-teaming platform for AI agents and moving my trading systems from backtest to live. I'd take something durable and used over something that demos well.
I work on the hard parts of the stack — security tooling, trading systems, and ML research. I care more about whether something gets used than whether it looks impressive, so most of what I build is aimed at running in the real world rather than sitting in a repo.