Jeet S. Swadia — Senior Data Scientist & AI Engineer

01 / Overview

At a Glance

With over 5 years of experience, I own the end-to-end delivery of high-impact AI and data science solutions for Fortune 500 clients in regulated healthcare and insurance environments.

My core focus is building Generative AI and agentic AI systems — with deep expertise in Azure platforms, LLM fine-tuning, and Responsible AI, ensuring fairness, transparency, and compliance.

I hold an M.S. in Electrical and Computer Engineering and a B.Tech. in Electrical Engineering, combining robust engineering fundamentals with strategic consulting capabilities to translate ambiguous business problems into measurable, AI-driven solutions.

See full work history →

5+

Years of Experience

J&J · MetLife · TCS

GenAI

Core Specialisation

Agentic · RAG · LLMs

Azure

Cloud Platform Expert

OpenAI · Synapse · DevOps

RAI

Responsible AI

Fairness · Explainability

02 / Capabilities

Tech Stack

Technologies I use to build production AI systems.

⌬ Generative AI & LLMs

Azure OpenAI OpenAI API Hugging Face LLM fine-tuning (LoRA/QLoRA) LangChain / LlamaIndex RAG Architectures Vector DBs (FAISS, Pinecone) Agentic AI Systems

◈ ML & Predictive Modeling

XGBoost LightGBM scikit-learn PyTorch / TensorFlow Bayesian Modeling A/B Experimentation Statistical Analysis

☁ Cloud & MLOps

Azure ML Azure DevOps / AKS AWS (S3, EC2, SageMaker) Docker / Kubernetes CI/CD (GitHub Actions) MLflow Model Monitoring

⟁ Data Engineering

PySpark Databricks Azure Synapse / ADF Delta Lake Airflow Large-scale ETL

{ } Languages

Python SQL Java SparkSQL

⚙ Consulting & Responsible AI

SHAP/LIME Explainability Fairness & Bias Assessment Strategic AI Advisory Executive Stakeholder Engagement Compliance Guardrails

Proficiency: Expert/Core Proficient

03 / Career

Professional Experience

Senior Data Scientist – GenAI, Agentic AI & Strategic Consulting

MetLife (IBSE Model, Team Gatekeepers)

May 2025 – Dec 2025

Owned end-to-end delivery of a high-impact AI solution for a Fortune 500 insurance client; collaborated with business stakeholders to define the problem, assessed risks, and delivered a production agentic AI system automating clinical document analysis for life insurance underwriting.
Designed and implemented Generative AI applications integrating Azure OpenAI, Hugging Face Transformers, and LangChain; applied prompt engineering, LoRA/QLoRA fine-tuning, and RAG architectures to build context-aware multimodal document understanding pipelines.
Built and deployed predictive and prescriptive ML models (XGBoost, LightGBM) with calibrated probability outputs; implemented SHAP/LIME-based explainability and compliance guardrails ensuring Responsible AI principles across fairness, transparency, and regulatory audit requirements.
Integrated RLHF/DPO human feedback loops to align LLM outputs with domain expert expectations; built evaluation harnesses tracking model accuracy, safety, latency, and cost across all production workflows.
Deployed production AI solutions on Azure (ML, OpenAI, AKS, DevOps) with full CI/CD, model monitoring, drift detection, and automated retraining; reduced manual underwriting effort by 40% through reliable, scalable AI delivery.

Generative AIAgentic AIAzure OpenAILangChainResponsible AI

Lead Data Scientist – Strategic AI Consulting

Johnson & Johnson (Voyager Metrics Portal)

May 2024 – Apr 2025

Led end-to-end delivery of an enterprise AI analytics platform for a Fortune 500 healthcare client as primary data science and consulting lead; defined business problem, designed solution architecture, developed project roadmap, and managed delivery across a 9-person cross-functional team.
Built and deployed scalable ML data pipelines on PySpark and Databricks integrating 50+ data sources; engineered systematic optimizations delivering 38% throughput improvement and maintained production reliability throughout.
Presented strategic AI recommendations, performance metrics, and business impact analyses to Finance Directors and C-suite stakeholders; acted as trusted advisor translating data science findings into strategic decisions.
Promoted Responsible AI practices: data governance, audit logging, access controls, and model transparency across all platform components; mentored engineers on ethical AI development standards.

PySparkDatabricksStrategic ConsultingData Governance

Senior Data Engineer – AI-Ready Data Infrastructure

Johnson & Johnson (Kenvue Platform Migration)

Aug 2022 – Feb 2024

Architected PySpark ETL pipelines migrating 40+ regulated healthcare datasets to Databricks Lakehouse; implemented data quality validation, schema governance, audit logging, and compliance controls supporting downstream ML model training.
Led cross-continent Agile team of 17+ engineers; enforced software engineering lifecycle best practices including version control, testing, code review, CI/CD, and production deployment standards.

PySparkDatabricks LakehouseETLCI/CD

Software Engineer

Tata Consultancy Services

Feb 2022 – Dec 2025

Built scalable Java and Oracle Database solutions; contributed to Agile cross-functional delivery teams.
Earned AWS Solutions Architect certification, expanding foundational cloud knowledge and system design capabilities.

JavaOracle DatabaseAWSAgile

Research Engineer – Applied AI & Statistical Modeling

Binghamton University

Jul 2020 – Jan 2022

Conducted applied research on Reinforcement Learning and Deep Neural Networks for autonomous systems on AWS; designed statistical experiments, built evaluation benchmarks, and communicated findings to academic and applied audiences.

Deep Neural NetworksReinforcement LearningAWS

04 / Credentials

Education & Certifications

01

Master of Science

Electrical & Computer Engineering

Binghamton University, NY

02

Bachelor of Technology

Electrical Engineering

Nirma University, Ahmedabad, India

03

Industry

Professional Certifications

Cloud & ML Architecture

05 / Work

Projects

★ Research

Real-time CNN Detection System

Built a real-time CNN-based detection system on video data. Designed robust data pipelines and comprehensive model evaluation frameworks, successfully achieving a 74% detection efficiency.

OpenCVTensorFlowPythonComputer Vision

★ Featured

Keno Optimizer — Lottery Analytics & Backtesting

A Streamlit dashboard plus experimental ML stack (LSTM, Random Forest, GPU-accelerated optimizer) for the MA State Lottery Keno. Reconstructs precise per-draw timestamps from a 400-draws-per-day rule, runs an ROI backtester against the official MA payout tables, and demonstrates honestly that Keno's certified RNG resists prediction.

// frequency_preview.py last 30d · 12,000 draws

Hot—

Cold—

ROI (8-spot, $2/draw)−24.7%

StreamlitPandasPlotlyPyTorchLSTMRandom Forest

06 / Writing

From the Blog

Technical deep-dives and engineering journals published on Medium.

ML Engineering Apr 2026 11 min read

I Built a Keno Optimizer. The Most Useful Result Was Proving It Couldn't Work.

A deep-dive into building a Streamlit ROI backtester, an LSTM sequence model, and a GPU-accelerated combinatorial search — and why the engineering matters more than the prediction.

Python LSTM Streamlit Random Forest GPU

Engineering Journal Apr 2026 7 min read

I Tried to Build a Trading Bot in One Night and Found Out How Hard "Just Connect to an API" Actually Is

A journal of building a Kalshi prediction-market trading bot — connecting REST APIs with RSA-PSS auth, running XGBoost on a local GPU, and learning why "just wire it up" is always a lie.

XGBoost REST API TimescaleDB Kelly Criterion Python

Let's Connect

I am based in Boston, MA and currently open to discussing new opportunities to build scalable, secure AI systems and drive strategic data initiatives.

Whether you're exploring a collaboration, looking for a technical leader in LLM architecture, or want to discuss Responsible AI implementations — my inbox is open.

swadiajeet@gmail.com

(838) 333-0802

or find me on

Engineering Intelligence. Scaling Insights.