Headline

AI Engineer | Agentic AI · RAG Pipelines · LLM Apps | Full-Stack

About

I build AI-powered apps end to end from the agentic backend to the React frontend. My focus is on practical LLM systems: tool-calling orchestration, RAG pipelines, long-term memory, MCP integrations, and cost-efficient architectures. I don't just wire up APIs I think about latency, cost, and what actually makes the UX feel right. In the past year I shipped four production apps: → NeutriAI an agentic nutrition coaching PWA with 15+ orchestrated tools, semantic long-term memory (pgvector + NVIDIA NeMo embeddings), and live Swiggy food ordering filtered by your remaining macros for the day. → Synth News a news aggregation platform processing 1,000+ articles daily. Cut operational costs from $15/day to $5 through batch API processing and vector search replacing live API calls. → Audiobook Creator a Windows desktop app that converts PDFs into audiobooks with zero cloud dependency, including zero-shot voice cloning from a 10-second sample. → AI Chrome Extension a privacy-first browser tool running LLMs entirely locally via Ollama, with no data leaving the device. I'm actively looking for AI Engineer or Full-Stack roles in Bengaluru or remote. If you're building something interesting with LLMs, reach out. Core skills: Python | JavaScript | Next.js | React | Node.js | LangChain | LangGraph | RAG | pgvector | OpenAI API | Gemini | Anthropic API | MCP | Supabase | TypeScript

Location

Bengaluru

AI stack

cursor

langchain

n8n

openai

Work experience

Founder at Neutri AI, Jan 2026 - Present · 7 mos, Bengaluru, Karnataka, India, Building NeutriAI, a mobile-first AI nutrition assistant PWA from the ground up. Architected an agentic core with 15+ orchestrated tools handling meal logging, pantry management, recipe generation, and live Swiggy food ordering via MCP integration. Built long-term semantic memory using NVIDIA NeMo Retriever embeddings with cosine dedup, keeping chat latency under 2 seconds. Implemented OAuth 2.1 + PKCE with AES-256-GCM token storage for secure commerce writes.

AI Engineer at SynthNews, Jul 2025 - Present · 1 yr 1 mo, Bengaluru, Karnataka, India, Self Built project: Built Synth News from scratch an AI-powered news aggregation platform that processes 1,000+ articles daily from 15+ publishers. Architected a RAG pipeline using vector embeddings and semantic clustering to synthesize multi-source stories into single coherent reads. Engineered an automated ingestion pipeline (RSS webhooks + web scraping + batch embeddings) with 2–3 hour publication latency. Built a hybrid search system combining semantic vector search with external news APIs, cutting monthly search costs 95% ($1,380 vs $10,500 for a pure API approach). Reduced overall operational costs from $15/day to $5/day through batch API processing and vector similarity search.

Education

Master of Computer Applications - MCA, Computer Science, Babasaheb Bhimrao Ambedkar University (BBAU), Lucknow

Bachelor of Science - BS, Mathematics, Dr. Ram Manohar Lohia Avadh University