Aditya Paradkar

I'm a quant trader turned software engineer. I’ve worked in derivatives, risk and quantitative research before moving to building the systems behind that work. I still trade my own capital on Indian exchanges, so I build for myself to put my money on the line.

That's shaped what I care about as an software engineer: systems that keep running when nobody's watching. Failure is treated gracefully, tests that actually mean backward compatibility, and operational tooling that makes 3am recoverable seem easy without losing sleep.

Statrex[https://aditya-paradkar.in/statrex] is where that shows most clearly — an end-to-end systematic trading system for Indian markets, live tick feed through to order placement and position management. Around 190k lines under 4,500 tests, on a ports-and-adapters architecture where brokers, market data, storage and even the clock sit behind explicit interfaces. That isn't architectural taste for its own sake: it's why three brokers can fail over without a strategy noticing which venue filled its order, and why the backtest-engine is bit-for-bit reproducible. It ships with a 65-command operational CLI, layered kill switches, and a storage layer handling retention, coverage auditing and incomplete-day recovery. Object-oriented and async throughout.

I have also done software consulting work. For Capital Ken (capitalken.com) I built a server giving AI assistants authenticated access to an investment-management database through nine structured tools — OAuth 2.1 with PKCE, RS256 JWTs, three-tier role-based access, rate limiting, circuit breakers with automatic recovery, and audit logging for easy error handling.

Alfred[https://aditya-paradkar/alfred], my self-hosted AI agent, runs 37 skills and 8 scheduled jobs across Telegram and WhatsApp behind a hardened permission gateway, where every skill pairs a natural-language instruction with a deterministic script and a confirmation gate on every write — the model decides what to ask, the script decides what gets written. Agentic and deterministic workflows, combined.

Public work you can actually read: mstat and xfer, Bash CLIs for homelab health monitoring and rsync transfers over Tailscale; an interactive system-design course built around a browser game where you build cloud infrastructure and survive traffic waves; and India-VIX1W, original research rebuilding NSE's India VIX methodology for weekly options.

I ship Python primarily, with Rust and Bash, on Linux and Docker.

I'm looking for fully remote mid-to-senior backend, data or platform where trading systems, financial data or AI tooling are the hard part. I have been self employed since 2024.