I've already shipped what your roadmap says.
Production RAG, fine-tuned open models and multimodal AI, with the full stack around them. Full-time roles first.
- Role
- GenAI / LLM engineer
- Where
- Remote · hybrid · Delhi NCR
- Intro call
- 20 min, Google Meet or phone
- 5production AI apps
- 14engineers led
- 2+years shipping AI
What I bring
Three things a team gets on day one, each already proven in production.
- 01
Production RAG that holds up
Hybrid dense and sparse retrieval, reranking, streaming citations and a confidence gate that refuses to answer on weak evidence. Shipped, not slideware.
- 02
Fine-tuning with honest evals
QLoRA through Transformers, PEFT and TRL, live training telemetry, and a side-by-side base-versus-tuned comparison before anything is called better.
- 03
The whole stack, end to end
FastAPI and Next.js, PostgreSQL and vector stores, Docker and CI/CD with logging and metrics from day one. Five services ran this way in production.
Proof of work
The five NextUpgrad systems, each with a case study. All 13 projects · Experience and stack on the résumé
RAG.NextUpgrad
2026Production RAG platform with hybrid dense + BM25 retrieval, a hallucination-blocking confidence gate and multi-provider fallback.
FineTune Studio
2026End-to-end QLoRA fine-tuning platform with live loss streaming and base-vs-fine-tuned evaluation.
DocuLens AI
2026Groq-powered document intelligence — extracts, compares and compliance-checks invoices and contracts with schema-validated, deterministic scoring.
FaceVision
2026Browser-only face detection, recognition and liveness checks with ONNX Runtime Web — the backend stores embeddings, never images.
OCR & Speech Workspace
2026Mistral-powered workspace for page-batched PDF OCR, live speech-to-text and document-scoped RAG chat with citations.
Quick answers
What hiring teams ask before the call.
Is Pranjul Rathour open to GenAI or LLM engineering roles?
Yes. Hiring · hackathon judging · campus sessions · Remote / India / Delhi NCR. Full-time roles first; remote, hybrid or Delhi NCR. The first step is a 20-minute intro call booked at https://pranjulrathour.scult.in/book?for=hire.
What has Pranjul Rathour shipped in production?
5 production AI systems at NextUpgrad Web Solutions: RAG.NextUpgrad (hybrid dense + BM25 retrieval with a confidence gate), FineTune Studio (end-to-end QLoRA with live loss streaming), DocuLens AI (document intelligence), FaceVision (browser-only face recognition and liveness) and OCR & Speech Workspace, plus client products at SCULT INDIA. 13 projects in all, each with a case study at https://pranjulrathour.scult.in/projects.
What stack does Pranjul Rathour work in?
Python and TypeScript; FastAPI and Next.js; Hugging Face Transformers, PEFT and TRL for fine-tuning; LangChain and vector stores for retrieval; OpenAI, Anthropic, Google and open models (Llama, Mistral); PostgreSQL, Docker and CI/CD. The full list is on the résumé.
How does the hiring intro call work?
Intro call: 20 min, Google Meet or phone; Technical deep-dive: 45 min, Google Meet, screen shared; Portfolio walkthrough: 30 min, Google Meet. Book at https://pranjulrathour.scult.in/book?for=hire; the request arrives on WhatsApp and Pranjul confirms within a day. Role, team, stack and timelines are covered on the call.
Where is Pranjul Rathour based, and does he work remotely?
India (IST, UTC+5:30). He works remote, hybrid or from Delhi NCR, and takes calls Monday to Saturday, 11:00 to 20:00 IST.
Where is Pranjul Rathour's résumé?
Online at https://pranjulrathour.scult.in/resume, with the same words as the PDF, which downloads from https://pranjulrathour.scult.in/data/Pranjul_Rathour_Resume.pdf.
Twenty minutes to find out if it's a fit.
Role, team, stack and timelines. Google Meet or phone, Monday to Saturday, India time (IST).