hi, I'm

Soham Jagrit

I build LLM pipelines, AI evaluation frameworks, and intelligent systems at the intersection of data science and healthcare AI. MS in Data Science from UT Arlington, May 2026.

projects
MindFlow

AI patient co-pilot for psychiatric telehealth — real-time transcription, session summaries, and clinical note generation. Built at Legion Health hackathon, SF.

↗ Won Health AI Hackathon — The Atlas Network, San Francisco

Next.js 15 Claude API Deepgram Supabase pgvector Inngest
Mobius

Open-source resume tailoring app for students — parses job descriptions and rewrites LaTeX resumes to match using semantic gap analysis.

↗ Open source — built for students navigating competitive job markets

Python Streamlit Anthropic API Docling LaTeX
AI eval framework

End-to-end evaluation system for RAG pipelines — RAGAS metrics, LLM-as-judge scoring, A/B testing interface, and automated regression detection.

↗ Reduced manual eval time by ~60% in production at Delta Air Lines

RAGAS LangChain Pinecone AWS Python
NutriBot

Multimodal RAG chatbot for personalized nutrition guidance — food image recognition, dietary history retrieval, and meal plan generation.

↗ Multimodal pipeline with per-user RAG on AWS

RAG PyTorch Flask Pinecone AWS S3
experience
2025 — 2026
Data Science Intern
Delta Air Lines
  • Built LLM pipelines for document processing and semantic search at enterprise scale
  • Designed AI evaluation frameworks using LLM-as-judge scoring and RAGAS retrieval metrics
  • Shipped a production RAG system that reduced manual document review time by ~40%
  • Collaborated with cross-functional teams to deploy models via AWS SageMaker and Lambda
2024 — 2026
MS in Data Science
University of Texas at Arlington
  • Focus areas: LLM systems, multimodal RAG, ML evaluation, and healthcare AI
  • Built end-to-end ML pipelines with Docker, AWS SageMaker, and PyTorch
  • Coursework in deep learning, NLP, statistical inference, and cloud-scale data engineering
skills
languages
Python SQL R JavaScript
AI / ML
LangChain LangGraph PyTorch TensorFlow RAGAS XGBoost
vector & search
Pinecone pgvector Supabase
cloud & infra
AWS S3 SageMaker Lambda EC2 Docker
APIs & tools
Anthropic API Deepgram Inngest Streamlit Flask
frontend
Next.js React Tailwind CSS
contact

let's work together

Open to full-time roles in data science, AI/ML engineering, and LLM systems — especially in healthcare AI, fintech, and developer tooling. Based in the US, open to remote.

email me LinkedIn GitHub