I build backend-heavy products where distributed systems, AI orchestration, and product judgment all have to hold together.
Right now that looks like duSraBheja — Personal AI Second Brain & MCP Server and dataGenie — Multi-Agent Data Analytics Platform, plus client work that had to survive re…
A chat grounded in my own writing — projects, decisions, what I'm building now. Use it for the question the case studies can't answer in two clicks.
Distributed backend, AI orchestration, Amazon-scale ops — what kind of role lines up with what I want to build.
Tradeoffs, decisions, what's live versus what's still in-flight.
Why I left Amazon, what's worth shipping now, and where my time goes.
MCP, retrieval, multi-provider routing, when to stop decomposing.
The problem isn't capturing information — it's retrieving it when you need it and knowing what you know. duSraBheja runs on a single droplet, captures from Discord and six+ local sources, classifies and embeds everything, and exposes the result to Claude Code, Codex, and a public chat through MCP and HTTP.
Non-technical people shouldn't need to write SQL to explore their own data. dataGenie collapses the loop — ingest CSV / Excel / PDF / database, ask a question, get an interactive chart with a written explanation and confidence score, with the underlying query shown so you can verify.
Most barbershop websites fail on operations, not branding. The owner needed a system, not just a brochure. The result is a React 18 frontend with a brown/gold design system fronting a Node/Express booking platform with Stripe, JWT, Google OAuth, and a natural-language admin assistant gated by a metrics registry.
A Jersey City espresso bar with no web presence and no in-house technical capacity needed a real site, a real domain, and an operator who could hand it over and have it stay up. The result is a single-page vanilla HTML/CSS/JS site on its own droplet with a release-versioned deploy pipeline, atomic symlink cutover, 5-release rollback, and zero-downtime cert renewal.