Looking for a job

Pranav pant

@pranavpant

Message

I build production-grade agentic RAG systems for medical and legal decision support.

United States

Message

What I'm looking for

I'm looking to build production AI systems for high-stakes decision support, especially agentic RAG, privacy-aware NLP, and grounded workflows where evaluation, safety controls, and human escalation matter.

I've built production-grade AI systems for high-stakes medical and legal workflows, including the Brown Heart Assistant at Joshi Health Foundation and privacy-aware NLP pipelines at Rocket Lawyer.

At Brown Heart, I built a Python/FastAPI multi-agent medical RAG assistant with hybrid retrieval, citation-grounded answers, confidence-based refusals, and HITL escalation for clinician and appointment workflows. I deployed Azure retrieval services with 113 passing tests and LangSmith tracing for production monitoring and debugging.

At Rocket Lawyer, I processed 3,000+ legal Q&A pairs through PII anonymization, compliance validation, deduplication, and content-quality assessment for Rocket Copilot. I fine-tuned a BERT NER model on 600K+ PII-tagged samples, reduced redundancy by 59.5%, and improved assessment accuracy by 31%.

I'm also building self-correcting Graph-RAG and contract-risk platforms using LangGraph, Neo4j, FAISS, FastAPI, and NVIDIA-hosted LLMs. My research includes eight peer-reviewed publications and two Best Paper awards.

Experience

Work history, roles, and key accomplishments

Current

AI/Data Engineer

Current

Joshi Health Foundation (Brown Heart)

Mar 2026 - Present (7 months)

Built a Python/FastAPI Brown Heart Assistant for an NGO, delivering citation-grounded cardiovascular answers from curated FAQ and MASALA Study sources with streaming responses, source metadata, and medical-safety refusal behavior. Designed a router-orchestrated multi-agent RAG architecture and implemented hybrid retrieval with BM25, Azure PostgreSQL/pgvector, HNSWlib, FastEmbed/NVIDIA embeddings,

Python fastAPI OpenAI API Azure PostgreSQL Pgvector HNSWlib FastEmbed NVIDIA Embeddings BM25 Reciprocal Rank Fusion Langsmith LangGraph MCP HITL HIPAA

AI/Data Engineering Intern (ML/NLP Focus)

Rocket Lawyer

Jun 2025 - Aug 2025 (2 months)

Developed a modular NLP data-quality pipeline in Python on GCP for Rocket Lawyer's Ask a Lawyer dataset, processing 3,000+ Q&A pairs through PII anonymization, LLM-as-a-judge compliance validation, embedding-based deduplication, and content-quality assessment. Built a PII anonymization module by fine-tuning a BERT-based NER model and reduced data redundancy by 59.5%.

Python GCP Vertex AI BERT NER LLM As A Judge Embeddings Deduplication Anonymization NLP

Summer Intern

Oil and Natural Gas Corporation (ONGC)

May 2023 - Aug 2023 (3 months)

Built ML pipelines for intrusion detection, malware classification, and semi-supervised facies classification using CNNs, Apache Spark, and Bayesian Optimization. Developed scalable feature-engineering workflows for packet-stream and well-log data, including a CNN-based malware detector achieving 95%+ accuracy.

Python Apache Spark CNN Bayesian Optimization Machine Learning Intrusion Detection Malware Classification