
Full-stack & AI systems engineer building agents, local-first RAG, and production SaaS
I ship production SaaS, autonomous AI agents, and developer tooling that solve real problems, built for production.
Experience
- Architected and owned the frontend for Commercial & Industrial (CNI) and Supply Chain Finance (SCF) solar loan journeys, streamlining end-to-end workflows from application through credit underwriting to disbursement.
- Engineered an AI document intelligence service with a structured RAG pipeline for bill extraction.
- Optimized LLM inference and document pipelines while supporting 8,000-10,000 weekly calls at under 1% failure rate.
- Built a unified, multi-provider LLM service layer routing across Google Gemini, AWS Bedrock, and OpenAI for internal chatbots and business services.
- Developed an automated Tech-Ops pipeline turning WhatsApp and Gmail issues into tracked tickets with live status dispatch to Sales, and a resource-aware data warehouse job scheduler.
- Founded and architected Hortiprise, a nursery-management SaaS digitizing end-to-end horticulture operations: plant inventory, sowing batches, germination tracking, quotations, billing, sales, and dispatch.
- Engineered a multimodal document pipeline (Google Cloud Vision + Tesseract + LLMs) to extract structured data from messy real-world Indian invoices.
- Built RAG-powered nursery intelligence tools for plant health analysis, document analysis, and demand forecasting.
- Integrated interactive WhatsApp-based ordering and automated dispatch confirmations, no dedicated app required.
- Designed and deployed a custom infrastructure monitoring system (NMS) tracking GCP VMs, Docker containers, MySQL, and Qdrant to catch resource exhaustion before it caused downtime.
- Automated roughly half of previously manual document workflows and added automated restocking alerts to prevent inventory mismatches.
- Developed and maintained candidate search and indexing infrastructure on Apache Solr, indexing high-volume candidate profiles and recruitment metadata to power complex recruiter filtering.
- Architected a recruiter lead-generation and attribution engine with token-based link tracking, connecting inbound applications back to originating recruiters as actionable leads.
- Optimized background processing and recurring overnight data pipelines generating recruiter analytics, improving throughput and reliability.
- Delivered end-to-end full-stack features, database schema migrations, and rapid production bug triaging.
- Built and maintained the React web platform for an Australian enterprise SMS gateway: campaign creation, messaging broadcasts, and contact lifecycle management.
- Spearheaded the core dashboard modernization from V1 to V2, redesigning UI architecture into reusable React components and pages.
- Engineered responsive, mobile-optimized interfaces across the messaging suite.
- Collaborated with backend engineers on automated two-way messaging workflows triggered by real-time replies and scheduled campaigns.
- Developed and maintained the company's public-facing React web apps and marketing properties.
Selected work
Hortiprise
Complete nursery-management SaaS with AI operations
End-to-end nursery lifecycle tracking from seed germination to dispatch; automated invoice/bill digitization via Google Cloud Vision OCR with Tesseract fallback; conversational WhatsApp ordering and support; background billing/subscription workers; custom infrastructure monitoring.
Real commercial production SaaS replacing pen-and-paper operations for plant nurseries, with 1,500+ commits across full-stack repositories.
DiffCommit AI
Published VS Code extension for semantic git commit generation
Analyzes staged git diffs and generates structured commit messages inside VS Code's source control panel. Supports 9 AI providers: Claude, GPT, Gemini, DeepSeek, xAI Grok, Mistral, Groq, OpenRouter, and local offline Ollama. Zero runtime dependencies.
Published on the Visual Studio Marketplace. Keeps API keys in the OS keychain via VS Code's native SecretStorage API, not plaintext config.
Syntra
Local-first codebase RAG assistant with incremental indexing
Indexes private codebases and docs for contextual AI Q&A. Smart incremental indexing inspects git diffs and content hashes to only re-embed modified chunks, cutting indexing overhead by 90%+.
Privacy-first: code never leaves the local machine. Eliminates re-index latency on active feature branches.
AI Investment Pipeline
Deterministic 3-stage startup triage & memo generation engine
Autonomous pipeline (source → analyze → recommend) for seed-stage VC thesis matching. Scores startups 0-100 against an AI infrastructure thesis and applies a deterministic rubric for Pass/Watch/Meeting verdicts, with on-disk JSON caching for zero-cost replay.
Eliminates LLM hallucination and decision drift in financial analysis: auditable, backed by a full Pytest suite and a decision log.
More projects
Skills
Education
Certifications
Beyond the code
Get in touch
Senior AI / full-stack engineer roles, technical collaborations, and select consulting. Gurugram / Delhi NCR, India.