Role: AI & Data Engineer - production LLM systems, solo, end to end
Focus: RAG · local/private AI · agentic AI (MCP) · OCR & data automation
Record: 2+ yrs shipping production AI · 5 enterprise platforms built solo · 100+ daily users · 200K+ transactions/month
Creed: 100% on-premise - client data never leaves client infrastructure
Security: Multi-tenant RBAC · guardrails · prompt-injection filtering
Base: Gurugram, India · remote & hybrid friendly
Now: Evolving Stratum & Excellia · LLM evals · agent orchestration
| System | Mission | Stack | Status |
|---|---|---|---|
Enterprise RAG |
Answers strictly from your documents - hybrid retrieval → RRF → cross-encoder → MMR → grounded refusal; tenant isolation in SQL | Python · FastAPI · pgvector · Ollama | |
Data intelligence |
Air-gapped spreadsheet intelligence - 500K-row files, evidence-backed answers; 19-tool MCP server + Excel add-in + web + API | Python · pandas · scikit-learn · MCP | |
Chrome Store |
Fills any web form via local LLM, zero cloud calls - 9-layer field-intent engine; verify-then-write, never auto-submit | JS · Chrome MV3 · Ollama | |
Document pipeline |
Multi-doc OCR + ML/LLM identity matching (Aadhaar, PAN, GST) - used in real Paytm & OYO audits | Python · Google Vision · Tesseract |
Closed-source @ TRPW: Limestone (bank reconciliation - 10K+ tx/month, 32 hrs/week saved) · Betel TMS · Betel VMS · SOP Intelligence Portal
Open to AI / Data Engineering roles & consulting - RAG systems, document automation, data platforms.
Gurugram, India ·
anuragsingh2445@gmail.com ·
linkedin.com/in/anurag2050

