Selected work
I build software I actually want to understand — clearing and settlement infrastructure, market tooling, virtualisation, data platforms, and the web. Mostly Rust, Python and C.
I lean toward systems that are deterministic, honest about their scope, and built in public. Most pages here link straight to the source, and I’m open to interesting work.
- 01 Open Clearing Engine An NSCCL-grade clearing and settlement engine in Rust — novation, multilateral netting, SPAN portfolio margin, event-sourced and byte-for-byte replayable. Rust · Fintech · Event sourcing 2026
- 02 Indie Market Analyst A local-first Indian-market toolkit: deterministic intraday scanner, dashboards, a backtester carrying the full Indian cost model, and an optional AI chat. Python · React · NSE / BSE 2026
- 03 Alt-data platform Point-in-time alternative-data platform for Indian equities. Medallion architecture, a feature store that is structurally anti-lookahead, plug-in sources. Python · TimescaleDB · Prefect 2026
- 04 VMManager QEMU/KVM machines and Incus/LXD containers in one native GTK4 window. Written in C against libvirt, ships as a single binary with no runtime to install. C · GTK4 · Virtualisation 2026
- 05 AiSoul A local-first personal AI harness for Android in Kotlin — the model, the memory and the tools all stay on the device. Kotlin · Compose · AI agent 2026
- 06 LPU Labs On-demand development environments a student can open in a browser tab. Container orchestration hidden behind a single button. FastAPI · React · Docker 2025
- 07 RLER A reinforcement-learning paper trader for NSE equities on 15-minute candles: tabular Q-learning, walk-forward validation, a Yew dashboard. Paper-only, on purpose. Rust · Reinforcement learning 2026
- 08 CS Archive A free, open-source computer-science knowledge hub across 20+ curated categories. Frozen at v3 and kept live as a reference snapshot. HTML · Knowledge base 2025