I take on freelance projects, contract engagements, and AI consulting for teams that need production-ready GenAI systems — not another proof-of-concept that dies in a Notion doc.
|
Sprint Projects 1–4 weeks POC · MVP · Prototype |
Contract Work 3–6 months 0-to-1 delivery |
AI Consulting Discovery · Scoping Architecture reviews |
Retainer Monthly partnership Fractional AI lead |
I'm a Forward Deployed AI Engineer with 2+ years shipping production LLM systems end-to-end. I don't just build demos — I embed with teams, do discovery, architect the thing, deliver it, and iterate post-launch.
- 🧠 12+ live systems across eCommerce, HRtech, fintech, and real-estate
- 🎯 Deep focus on RAG, LangGraph multi-agent orchestration, agentic AI, LLM eval
- 📈 Outcomes shipped: −40% recruiter workload · −35% API latency · −50% manual work · 99%+ uptime
- ✍️ Working style: direct, technical, results-focused — you'll get a shipping thing, not a slide deck
- 🌏 Based in Indore, India — collaborate with teams anywhere in the world
- 🎧 Off-hours: coffee, music, and picking apart new AI papers
AI / LLM specialty stack:
OpenAI · Anthropic · Google Vertex AI · LangChain · LangGraph · Hugging Face · Pinecone · FAISS · ChromaDB
|
Production hybrid RAG library. Pluggable chunkers, embedders (BGE-M3, OpenAI), vector stores (Qdrant, in-memory), and cross-encoder reranking. Async-native, strict-typed.
|
Clone-and-run multi-agent template. Router + specialists, tool loop, streaming, iteration cap. Fake-model tests so CI runs without API keys.
|
|
Multi-agent customer-support triage: classify → specialist → quality-gate → auto-send or escalate. FastAPI webhook. Production shape, not a demo.
|
LLM-powered invoice PDF → structured JSON. FastAPI service + CLI. Pydantic-validated schema with auto-reconcile guardrail.
|
|
3D animated portfolio in Next.js 14 + React Three Fiber + Framer Motion. Distort-material hero, scroll-linked nav, glassmorphic sections.
|
GitHub Action for LLM-powered PR review. Inline severity comments + APPROVE / COMMENT / REQUEST_CHANGES verdict. Supports OpenAI + Anthropic.
|
|
Small, dependency-light RAG evaluation. hit@k, MRR, NDCG, precision, recall + LLM faithfulness judge. Zero framework baggage.
|
Contributor with 280+ commits to Sparrow — next-gen API testing platform. Backend features, Sentry logging, feature dev at Techdome.
|
Most client work is under NDA — detailed case studies live on my portfolio.
| Public repos | OSS libraries shipped | Tests across projects | Commits to Sparrow OSS |
Top languages (by lines shipped this year): Python · TypeScript · JavaScript · Jupyter · HTML · CSS
Looking to ship a production AI system? Need someone who can scope, architect, and deliver — not just tune a prompt for a Loom demo? I'm your person.
📩 prayas1711@gmail.com · 💼 LinkedIn · 🌐 Portfolio · 🐦 X
💡 Deployed where the problem lives. The "forward deployed" part isn't just a title.