I've been building software for real clients since I was 19, mostly the systems they use to run their business day to day. I like taking a problem from nothing to shipped and owning the whole thing, the database, the backend, the frontend, and the AI parts where they earn their place.

B.Tech AI & ML, Symbiosis Institute of Technology, Pune · 50+ systems shipped for 15+ clients

I take a business problem and build the whole thing end to end, then keep it running. Most of my work is for clients in family entertainment, edtech, manufacturing, and consulting. I'm at my best when I own a system start to finish and get to make the calls myself.

• AI agents and automation: multi-step workflows, tool calling, RAG, human approval gates, n8n and custom pipelines
• LLM integration: Claude, OpenAI, Gemini, and open source models like Llama and GPT-OSS, with model routing, structured outputs, and fallbacks
• Operations dashboards: role based, Postgres backed, wired to third party APIs, built so owners can actually run their business
• Computer vision: verification pipelines, face recognition with InsightFace and FAISS

AI-Verified Operations System. Field staff submit photo and video evidence of their daily work, and a dual model pipeline (Llama 4 Scout plus Claude Vision) checks each one, auto approves about 95%, and sends only the uncertain ones to a human. Over 30,850 checks in production.

Revenue Intelligence Dashboard. Reads all of a venue's sales and operations data and tells the owner, in plain words, what's making money and what isn't. Role based views for crew, managers, and ownership.

Multichannel Sales Pipeline. Enquiry to close automation across email, WhatsApp, and an automated AI voice agent that qualifies leads and hands off to a human, with transcription, approval gates, and follow up loops.

Secure Video Platform. Device locked course delivery for edtech institutes so paid content can't be screen recorded, screenshotted, or shared. Cloud to offline encryption, seat based licensing, two desktop apps and a web dashboard.

Offline Manufacturing Dashboard. A full operations system for a diamond manufacturer that runs completely offline: order tracking, live inventory by weight and type, worker performance, and outstanding payments.

Face Recognition Entry. Cardless member entry recognizing 1,000+ members with InsightFace and FAISS, plus 50,000+ digital waivers.

Backend: Python, FastAPI, Flask Frontend: Next.js, React, TypeScript, vanilla JS Data: PostgreSQL, Supabase, Redis, MongoDB AI: Claude, OpenAI, Gemini, open source LLMs, RAG, FAISS, InsightFace Automation: n8n Infra: Docker, Git