I build production AI systems end-to-end, from data pipelines to LLM-powered products to deployment and monitoring.
What I build (in plain English)
A few highlights that reflect how I work: measurable outcomes, production constraints, and shipping.
More detail and specifics live inside the case studies below.
Product-minded builds with case studies, tradeoffs, and outcomes.
Client -> proxy -> router -> provider adapters, with auth, rate limits, and cache.
Unified proxy server for multiple LLM providers with one consistent API.
Auth, rate limiting, streaming, monitoring, and provider health checks behind one gateway.
Answering with citations and source links.
AI-powered support agent for an API product (Next.js + FastAPI).
Interactive docs browsing, real-time chat, Supabase auth, and admin tooling.
Agenda view with quick actions and cleaner layout.
Maintained fork of a KDE Plasma Event Calendar widget focused on daily usability.
Google Calendar sync fixes, quality-of-life improvements, and ongoing Plasma 6 support.
Train -> evaluate -> register -> deploy.
Production-style CI/CD for an ML API with CML, Docker, Kubernetes, and MLflow.
Automated testing, container builds, deployment patterns, and model lifecycle thinking.
Principles I use to ship reliable systems without overbuilding the first version.
I optimize for outcomes and evidence: demos, docs, tests, and clear decisions.
Auth, rate limits, retries, and observability are part of the product, not extras.
Clean interfaces and adapters keep systems flexible as requirements evolve.
I work across AI/ML and software engineering, with a focus on systems that ship: APIs, pipelines, deployment, and the product layer users interact with.
I'm especially interested in education tech, healthcare tech, and applied AI where reliability and clarity matter.
Outside of work, I will happily fix the annoying edge case in an open-source tool because it makes someone's daily workflow smoother.
Now
Building reliable AI systems with tight feedback loops.
Location: Bengaluru, Karnataka, India
Focus: Retrieval eval, routing policies, and deployment hygiene.
Personal: KDE + Linux customization, small tools that improve daily flow.
When I learn something useful, I write it down.
Notes and experiments from the IITM GenAI track.
A lightweight evaluation loop for retrieval quality and answer grounding.
Guardrails and policy choices that keep routing stable under real traffic.