I'm the entire engineering department at Nyayanidhi, a legal-tech startup building AI-powered document processing for Indian law firms. Every line of backend code is mine.
The platform runs on FastAPI, async SQLAlchemy, and PostgreSQL, deployed on GCP Cloud Run. I built a 7-phase document ingest pipeline that processes 500+ PDFs in under 10 minutes using fan-in/fan-out orchestration via Cloud Workflows. Each phase is independently restartable with checkpoint-based resumption.
The multi-agent drafting system generates contracts from 19 variant templates using schema-driven LLM tool creation and multi-phase blackboard coordination. I built, evaluated, and killed features when they didn't earn their complexity — including a condenser agent I recognised as over-engineered and removed.
On the LLM infrastructure side: prompt caching for Anthropic and Gemini providers, multi-modal content support, provider migration (Gemini to Sonnet for tool-use reliability), and connection pool tuning that cut p95 API latency by 40%.
At the Leibniz Supercomputing Centre in Munich, I worked on HPC cluster management for systems with 1000+ compute nodes connected via InfiniBand. Built automation tooling for patch workflows and monitoring, profiled system performance with perf, Intel VTune, and Valgrind, and integrated everything with SLURM job scheduling.
This was bare metal systems engineering — no cloud abstractions, no managed services. NUMA topology, cache hierarchies, MPI, OpenMP. The gap between "cloud engineer" and "systems engineer" is real, and I've worked on both sides.
During the COVID pandemic, I helped scale Money Network infrastructure for the US Treasury's Economic Impact Program — delivering stimulus debit cards to 3 million Americans. Government deadlines, zero tolerance for downtime, real consequences for failure.
I also rebuilt the microservices backend for Money Network's customer care system (Spring Boot, supporting 2000+ programs) and re-architected the disputes platform handling 1M+ daily transactions.
Before all of this, I contributed to GNOME Music — the default music player for the GNOME desktop environment. 14+ commits to the upstream project. Open source was how I learned to write code that other people have to read.
MSc Computer Science at TU München (coursework in HPC, parallel computing, systems programming). B.Tech Computer Science from Amrita School of Engineering, Tamil Nadu.
I speak English, German, Tamil, Telugu, Hindi, Kannada, and Malayalam.