Data / AI / Healthcare / Agentic Systems

Building intelligent pipelines that turn complex data into trusted decisions.

I design agentic AI systems, retrieval workflows, predictive models, and production data platforms for healthcare and analytics teams, from messy source material to reliable downstream action.

7+ years across data, AI, and product engineering.

Agentic AI pipelinesRAG and retrieval systemsLLM-as-a-judge evaluationHealthcare data workflowsSnowflake + AWS + PythonLangChain / LangGraph / OpenAI
Deepanshu Gupta
Current focusAgentic AI for data workflows
Production stackSnowflake, AWS ECS, Python, OpenAI
Domain edgeHealthcare, life sciences, and commercial data

Expertise

Applied AI, data systems, and healthcare analytics.

Work centered on agentic pipelines, retrieval, evaluation, and production data delivery, with a consistent focus on useful outcomes in complex domains.

Agentic pipeline design

I build multi-step systems that retrieve context, call tools, validate outputs, and hand structured results back to real teams.

LangChainLangGraphOpenAIStructured outputs

Healthcare AI systems

My strongest domain is healthcare and life sciences: lab reports, claims, EHR, diagnostic signals, and precision medicine data products.

Lab reportsClaimsEHRPrecision medicine

Retrieval, eval, and trust

I care about grounded answers, LLM-as-a-judge evaluation, quality control, and systems that can be trusted outside the notebook.

RAGLLM evalGuardrailsQuality control

Production data platforms

I ship with Snowflake, AWS, ECS, SQL, and Python so the AI layer sits on top of data plumbing that is actually usable in production.

SnowflakeAWS ECSSQLPython

Experience

Career chapters shaped by systems, models, and delivery.

My path moves from enterprise product engineering into healthcare data science and now toward agentic AI delivery, where retrieval, evaluation, and production constraints all matter at once.

01
DataClaims, EHR, lab reports, commercial intelligence
02
RetrieveSearch, embed, normalize, and ground
03
ReasonAgent workflows, tool use, structured JSON outputs
04
EvaluateLLM-as-a-judge, QA loops, traceability
05
DeployAWS ECS, Snowflake, reporting, feedback loops
01Dec 2023 - Present

Current chapter

Data Scientist

Diaceutics PLC / Belfast, UK

Building production-grade agentic AI systems for healthcare data workflows, from extraction and retrieval to evaluation, reporting, and delivery.

Impact

  • Built and deployed end-to-end LLM workflows across lab reports, claims, EHR, and commercial healthcare datasets.
  • Delivered agentic pipelines with Python, LangChain, LangGraph, OpenAI, AWS ECS, and Snowflake-backed outputs.
  • Reduced execution cost by about 60 percent while increasing cleaned-data volume by about 40 percent.
  • Introduced LLM-as-a-judge quality control and structured JSON outputs for higher trust and better downstream usability.

Systems used

PythonLangChainLangGraphOpenAIAWS ECSSnowflakeSQLRAG
Representative work

Data Extraction AI Pipeline / Sales AI Pipeline / Diagnosis Events Detection / Affiliation Name RAG / Geneview

02Jul 2023 - Sep 2023

Modeling chapter

Data Analytics Intern

Diaceutics PLC / Belfast, UK

Worked on exploratory analysis, predictive modelling, and healthcare outcome analysis with a mix of classical ML and sequence models.

Impact

  • Ran exploratory analysis and text-heavy healthcare data investigation for treatment and patient-outcome questions.
  • Built models using Random Forest, XGBoost, AdaBoost, and bidirectional LSTM workflows.
  • Achieved roughly 75 percent accuracy in treatment outcome analysis experiments.

Systems used

PythonXGBoostRandom ForestBiLSTMTensorFlowEDAHealthcare analytics
Representative work

Treatment outcome analysis / Predictive modelling / Patient record exploration

03Feb 2019 - Aug 2022

Product systems chapter

Product Engineer

Tata Consultancy Services - Digitate / Pune, India

Built enterprise AI-ops and AI-assurance product features, with a strong grounding in shipping software, monitoring systems, and operational reliability.

Impact

  • Delivered product features for AI-driven IT operations, incident management, monitoring, and proactive issue resolution.
  • Worked across JavaScript, Angular, Java, PostgreSQL, Python, Docker, Jenkins, Ansible, and HAProxy.
  • Received the Star Employee Award and won first prize at the TCS Ideathon.

Systems used

AngularJavaScriptJavaPostgreSQLDockerJenkinsAnsibleAIOps
Representative work

AI assurance / AIOps workflows / Monitoring and resolution features

Education

Academic grounding

2022 - 2023

MSc in Data Analytics

Queen's University Belfast / Belfast, UK

Graduated with Commendation. Focused on machine learning, predictive modelling, applied analytics, and practical data science.

2014 - 2018

BTech in Mechanical Engineering

KIET Group of Institutions / India

Graduated First Class. Early research centered on supply-chain optimization and inventory decision methods.

Credentials

Certifications

Publications

Research roots and recognition

  • Star Employee Award, TCS Digitate
  • First Prize, TCS Ideathon
  • Technical writing on AI systems, reasoning models, and evaluation workflows on Medium

Writing

Field notes, technical writing, and longer-form thinking.

Writing that explores system design, evaluation, reasoning models, and the practical tradeoffs behind shipping AI work.

Loading your latest articles...

Contact

Let's talk about the system behind the AI.

Whether it is a production agent workflow, a retrieval-heavy use case, or a healthcare analytics problem that needs structure, I'm happy to compare notes.