Open to opportunities

Purvi User

@purviuser

Message

AI Engineer building production agentic RAG and ML systems end-to-end for enterprise impact.

India

Message

What I'm looking for

I’m looking for a role where I can build and operate production GenAI/ML systems—especially agentic RAG—owning evaluation, monitoring, and guardrails end-to-end, while partnering with stakeholders to deliver reliable, user-ready outcomes.

I’m an AI Engineer with production experience building, deploying, and operating ML and generative AI systems across the full lifecycle—from data pipeline design and model development to evaluation and monitoring. I focus on practical outcomes, turning complex AI trade-offs into actionable guidance for both clients and non-technical stakeholders.

I’ve led LangGraph-based agentic RAG pipelines and multi-agent orchestration in live settings, including prompt design and evaluation, retrieval tuning, structured output schemas, guardrails, and continuous monitoring. I’ve also operationalized an end-to-end computer vision pipeline (garment segmentation → classification → rendering → video synthesis), containerized with Docker and served via FastAPI with per-stage quality loops to prevent regressions.

My background spans computer vision, federated learning, and applied AI across healthcare, agriculture, and B2B intelligence, supported by research output (6 Springer papers) and a strong evaluation mindset. Projects like SoyBot and a multi-agent platform “CrisisGen” shaped how I benchmark reliability, observability, and guardrails to ship systems that users can trust.

Experience

Work history, roles, and key accomplishments

Current

AI Engineer

Current

Reqpedia

Oct 2025 - Present (8 months)

Designed and operated LangGraph-based agentic RAG pipelines for a live B2B intelligence platform used daily by enterprise sales teams. Built and productionalized an end-to-end computer vision workflow (segmentation→classification→rendering→video synthesis) with Docker/FastAPI and per-stage quality evaluation loops.

LangGraph Agentic RAG Prompt Engineering Structured Outputs GuardRails Docker fastAPI

AI/ML Engineer

ICAR - National Soybean Research Institute

May 2025 - Sep 2025 (4 months)

Designed, deployed, and operated SoyBot, a production LLM-powered multi-agent advisory system using a full LangGraph RAG pipeline for agricultural domain queries. Implemented evaluation and monitoring with golden query sets and hallucination/accuracy regression checks, and integrated speech-to-text plus multilingual translation for real end-users.

LangGraph Golden Set Evaluation Hallucination Rate Analysis Speech to Text Retrieval Evaluation

ML Research Associate

IIPS, Devi Ahilya Vishwavidyalaya

Jan 2024 - Jul 2025 (1 year 6 months)

Ran ML projects across data pipelines, model training, evaluation, and documentation, publishing 6 Springer papers across computer vision, federated learning, and agentic AI. Built a federated learning framework with Differential Privacy (TF Privacy ε=2.0) that achieved 93% diagnostic accuracy and benchmarked 6 segmentation models using per-class IoU on 1,564 images.

Federated Learning Segmentation Model Training Per Class IoU DenseNet121 ResNet50 Research & Documentation