Akash Gupta · Senior Data Scientist · GenAI Engineer
I design intelligent systems that ship to production.
5+ years across Generative AI, Machine Learning, and large‑scale data products. From clinical appeal‑letter copilots to graph‑powered policy chatbots, I turn ambiguous ideas into reliable, observable AI systems.
Experience
5+ years
Building data & GenAI systems at healthcare and BFSI scale.
Hackathons
65+ top ranks
Featured MachineHack Grandmaster & Kaggle Expert.
Focus
GenAI systems · RAG & knowledge graphs · LLMOps on Azure ML · production ML for healthcare & BFSI.
From Kaggle & MachineHack podiums to shipping GenAI copilots in production.
I design LLM systems as products: observable, debuggable, and safe by default.
Appeal-letter copilots. PolicyTech RAG chatbots. Next-best-action engines at healthcare scale.
The goal is simple: turn cutting-edge research into calm, reliable workflows.
Selected systems & stories
A snapshot of the GenAI, ML, and analytics work I've shipped across healthcare, BFSI, and competitive data science.
Appeal Letter Copilot
LLM-based assistant that drafts medical appeal letters from clinical history, aligning reasoning with CMS guidelines using RAG, fine-tuning, and nurse-in-the-loop validation.
Outcome
Reduced nurse drafting time from ~1 hour to minutes, freeing clinical staff for higher-value work.
2024 — Present · Ensemble Health Partners
PolicyTech RAG Chatbot
Low-latency RAG chatbot over internal policy documents, combining Azure AI Search, GPT-4/4o, and a Neo4j knowledge graph for multi-hop reasoning.
Outcome
Dramatically cut policy lookup time and enabled self-serve, always up-to-date guidance across teams.
2024 · Ensemble Health Partners
Next Best Action Prediction
Multi-class model trained on 75M+ rows and 20K+ features to recommend optimal account-resolution actions, with LLM-assisted feature engineering and client-specific retraining pipelines.
Outcome
Reached ~50% accuracy on the top 20% high-impact accounts, driving proactive interventions and revenue uplift.
2023 — 2024 · Ensemble Health Partners
Hackathon Wins & Research
1st place in challenges including insect species classification, dance-form recognition, and end-to-end ML systems; co-authored research on automated image captioning (ITBT-19).
Outcome
65+ podium finishes across platforms, constantly stress-testing modeling approaches and feature engineering ideas.
Ongoing · Kaggle & MachineHack
My Arsenal
A collection of the tools, frameworks, and technologies I use to build intelligent systems.
Generative AI & LLMs
Overview
Deep expertise in building agentic workflows, fine-tuning large language models, and implementing RAG architectures for production-grade AI systems.
Frameworks & Libraries
Overview
Proficient with the modern ML stack for training, evaluating, and deploying models, from classical machine learning to deep neural networks.
Cloud & MLOps
Overview
End-to-end MLOps capabilities including CI/CD pipelines, model monitoring, and cloud infrastructure management across major providers.
Programming & Core Tech
Overview
Solid engineering foundation rooted in Python, data science principles, and effective problem-solving strategies.
Path to Now
Data Scientist II · GenAI & ML
Ensemble Health Partners
Owning end-to-end GenAI and ML delivery: architecture, RAG/LLM design, Azure ML deployment, monitoring, and scale for enterprise healthcare workflows.
Asst. Manager · Data Scientist I
HSBC
Built and deployed ML/NLP pipelines for banking analytics, including NPS driver analysis and agent workspace recommendations to guide service improvements.
Data Analyst
Tata Consultancy Services
Developed time-series forecasting and supply-demand models, turning noisy operational data into actionable business signals.
Let's design the next experiment.
Whether it's a GenAI copilot, a production ML system, or a high-stakes data problem, I'm always excited to collaborate on deeply technical, high‑impact work.
it1akgec@gmail.com