Snehit Vaddi

AI Researcher

United States3 yrs 4 mos experience

Most Likely To SwitchAI ML Practitioner

Key Highlights

Developed AI solutions saving physicians 2 hours daily.
Published research on bacterial contamination detection.
Built cost-effective in-house document processing system.

Stackforce AI infers this person is a Healthcare AI Engineer with expertise in machine learning and data engineering.

Contact

Skills

Core Skills

AiData EngineeringMachine Learning

Other Skills

LangChainOpenAI Agents SDKVision Language ModelsData ProcessingDocument ManagementPyTorchComputer VisionYOLOv8GradCAMApache SparkNatural Language Processing (NLP)Azure DatabricksPredictive ModelingAWS SageMakerCUDA

About

⚡ Building ModMed AI Scribe - processing 500M+ patient conversations in real-time, Saving physicians 2 hours daily, 206% ROI, HIPAA-compliant, 40K+ providers across 11 specialties. 🎓 Graduated with a Master's in CS from the University of Florida. 💡 Published research on detecting bacterial contamination prior to visible degradation using hyperspectral imaging and vision transformers. ⚡Exploring the exciting intersections of Big Data, ML & GenAI. 🧠 Love working with Startups, an contributing to opensource. 🏔 In my free time, you can find me organizing events, listening to podcasts, and travelling. 💪🏽 Strengths: SQL & NoSQL I AI, ML, CV | Data Science |

Experience

3 yrs 4 mos

Total Experience

1 yr 1 mo

Average Tenure

1 yr 3 mos

Current Experience

Modmed

AI Engineer

Feb 2025 – Present · 1 yr 3 mos · United States

Shipped ModMed Scribe that automates clinical documentation for 15000+ of providers across 11+ specialties.
Designed & Developed "Agentic Docuemnt Processing" to replace a third-party document processing vendor with an in-house multi-agent system using Vision Language Models, in numbers $400K to $20K, cutting spend by 95%, currently handling millions of pages monthly.
On the data side, built a semantic layer over a fragmented EHR warehouse so business users and AI agents can query clinical metrics in plain English instead of waiting on analyst queues.
Also developed and open-sourced MEDHALT, a medical hallucination detection framework now integrated into production for real-time clinical note validation.
Day-to-day: LangChain, OpenAI Agents SDK, VLMs, fine-tuned Whisper, DeBERTa, Databricks.

LangChainOpenAI Agents SDKVision Language ModelsData EngineeringAI

University of florida

Graduate Research Assistant - AI/ML

Jun 2024 – Dec 2024 · 6 mos · Gainesville, FL · Hybrid

R&D on explainable AI for agriculture and food safety. Architected end-to-end pipeline for E. coli detection using computer vision - YOLOv8 + ViTs with GradCAM explainability (92% accuracy). Processed 100GB+ agricultural imagery and spectral data.
Research published in SPIE, ASBE and IEEE conferences.

PyTorchData EngineeringComputer VisionAIMachine Learning

Stealth startup

AI Software Engineer Intern

May 2024 – Jul 2024 · 2 mos · United States · Remote

Engineered AI search and support systems using RAG/GenAI, optimizing document retrieval across 50k+ documents. Designed Ingestion, ETL pipelines and retrieval pipelines to enable low-latency, high-precision query resolution.

At&t

Software Data Engineer

Jun 2021 – Jan 2023 · 1 yr 7 mos · India

Built ML & Data pipelines on Azure Databricks for technician dispatch optimization. Developed predictive models analyzing service patterns, reducing unnecessary dispatches by 12K annually - $2M cost savings. (Associated with Accenture)

Apache SparkNatural Language Processing (NLP)Data EngineeringMachine Learning