Connect in M
Agentic AI Impact Experience Writing Skills Hire Me
Senior Data Scientist · AI/ML Leader · US B1/B2 Visa

Turning Data
Into Business
Revenue.

10 years transforming raw data into measurable outcomes — from reducing churn at Vodafone to driving campaign ROI at Home Depot. I build AI systems that pay for themselves.

10+ Years in DS/AI
6 Enterprises
5+ CX Domains
What I Bring to Your Team
Agentic AI
Multi-Agent Systems with LangGraph
Building production-grade agentic workflows — multi-step reasoning, tool use, and memory — using LangGraph, LangChain, and OpenAI.
Revenue Impact
Predictive Models That Convert
Lookalike and propensity models at Office Depot directly influenced AOV uplift by surfacing high-intent customers for monthly campaigns.
GenAI & LLMs
AI Applications That Actually Work
Text2SQL, RAG pipelines, and LangChain agents in production — tested across GPT-4, LLaMA, and domain-specific models.
End-to-End Ownership
From POC to Production
Led requirement gathering, built POCs, and deployed at scale — including computer vision systems that replaced manual QA at Mercedes-Benz R&D.
Frontier Expertise

Agentic AI &
LangGraph Systems

I architect multi-agent AI systems that don't just respond — they plan, reason, act, and adapt. Moving beyond single-shot LLM calls to orchestrated pipelines where agents collaborate, use tools, maintain memory, and recover from failures autonomously.

🔗
LangGraph Orchestration
Stateful, cyclical agent graphs with conditional edges, human-in-the-loop checkpoints, and persistent memory across multi-turn reasoning chains.
🧠
Multi-Model Agent Architecture
Routing tasks across specialised models — GPT-4 for reasoning, LLaMA for domain knowledge, fine-tuned models for structured output — within a single agentic pipeline.
🛠️
Tool-Using Agents in Production
Agents equipped with SQL executors, web search, vector retrieval, API calls, and code interpreters — deployed beyond prototype into real business workflows.
📊
RAG + Agentic Retrieval
Hybrid retrieval pipelines combining dense vector search (FAISS/Pinecone) with structured query generation — giving agents grounded, accurate, up-to-date knowledge.
Featured Agentic Projects
Multi-Agent System
🧠 Multi-Agent Counseling System
A coordinated multi-agent framework where specialised agents handle intake, assessment, and guidance — with shared state managed via LangGraph's stateful graph execution.
LangGraphOpenAIMemoryMulti-Agent
View on GitHub →
Agentic Workflow
🔗 LangGraph Agentic Workflow
Production-grade agentic workflow engine — cyclical graph with conditional routing, tool-call nodes, retry logic, and structured output parsing for downstream systems.
LangGraphLangChainTool UseConditional Edges
View on GitHub →
GenAI Application
💬 AI Life Coach
Persistent memory-backed coaching agent that tracks user goals across sessions, uses retrieval-augmented context, and adapts responses based on longitudinal conversation history.
OpenAIRAGPersistent MemoryGenAI
View on GitHub →
Computer Vision + AI
🎥 Shoplifting Detection
Real-time YOLO-based detection system with event-driven alerting — designed for low-latency inference at the edge with configurable confidence thresholds.
YOLOPyTorchOpenCVReal-time
View on GitHub →
View All Projects on GitHub →
Measurable Results

Business Value I've Delivered

Every model I build is anchored to a business outcome — not just accuracy metrics.

🎯
↑ AOV
Average Order Value Growth
Propensity and lookalike models at Office Depot surfaced high-intent customers for monthly campaigns, directly influencing AOV uplift.
📈
↓ CAC
Reduced Customer Acquisition Cost
Incrementality frameworks at Home Depot eliminated wasted spend and proved true ROI on new feature launches.
🏭
↓ QA Cost
Manufacturing Quality ROI
YOLO-V5 defect detection at Mercedes-Benz R&D replaced manual inspection, reducing workforce costs and improving catch rates.
Real-time
Streaming ML Pipelines
Apache Kafka-based real-time ML at Tata Technologies enabling instant sensor failure detection and predictive maintenance at scale.
💬
NLP → SQL
Democratised Data Access
Text2SQL & RAG at MarketsMojo put equity insights into plain English — reducing analyst dependency and accelerating decision cycles.
📡
Multi-region
Scalable Telecom AI at Vodafone
Transactional behaviour models deployed across regions at VOIS India — powering Finance and CX decisions for millions of customers.
Career Timeline

10 Years, 6 Enterprises,
Consistent Delivery

A track record across telecom, retail, automotive, and fintech — always focused on business outcomes.

VOIS India — Vodafone Intelligent Solutions
Lead Data Scientist · Manager
Dec 2024 — Present
  • Leading the Advanced Analytics & AI Tribe, delivering AI-driven insights across Finance and Telecom domains for millions of customers across multiple regions.
  • Building transactional behaviour models identifying spending patterns, risk indicators, and campaign responsiveness — directly informing commercial strategy.
  • End-to-end ML workflow ownership: training, deployment via Iguazio + Docker, monitoring — ensuring production reliability and scalability.
  • Designing AWS QuickSight executive dashboards translating model outputs into actionable business metrics for senior leadership.
PythonSQLOracleIguazioDockerAWS QuickSightScikit-learn
MarketsMojo — Fintech Research
Data Scientist · GenAI
Jun 2024 — Dec 2024
  • Built production Text2SQL + Text2NoSQL + RAG application enabling non-technical analysts to query equity data in plain English.
  • Benchmarked GPT-4, LLaMA, and Palmyra Finance models for domain-specific financial language understanding.
  • Implemented Monte Carlo simulation for portfolio diversification analysis, delivering country-level risk insights to investment teams.
LangChainGPT-4LLaMARAGMongoDBStreamlit
Home Depot — Findability Sciences
Senior Data Scientist
Dec 2022 — Mar 2024
  • Owned end-to-end KPI monitoring (visits, revenue, AOV, CTR) for a high-traffic e-commerce platform, identifying seasonal demand cycles.
  • Built Prophet-based forecasting models capturing trend and seasonality, enabling accurate demand planning and reducing stockouts.
  • Designed an incrementality measurement framework to quantify true revenue lift of campaigns and new features.
BigQueryVertex AIProphetPythonTableau
Office Depot — CompuCom CSI Ltd
Data Scientist · CX & Marketing
Dec 2020 — Dec 2022
  • Developed propensity-based lookalike models powering monthly targeted campaigns and driving measurable Sales and AOV uplift.
  • Architected a multi-platform feature store aggregating data to train higher-accuracy behavioural models.
  • XGBoost ensemble models outperformed baselines; recommendations consumed directly by the SAS marketing team.
SnowflakeAWS SageMakerXGBoostRandom ForestTableau
Tata Technologies Ltd.
ML Engineer · IoT & Streaming
Mar 2019 — Dec 2020
  • Engineered a real-time ML pipeline using Apache Kafka for streaming sensor data, enabling instant anomaly detection.
  • Applied unsupervised learning to uncover hidden failure patterns in unlabelled sensor data, reducing unplanned downtime.
Apache KafkaMongoDBDockerAWS LambdaSVM
Mercedes-Benz R&D India
Computer Vision Engineer
Jun 2017 — Mar 2019
  • Delivered a production-grade defect detection system using YOLO-V5 replacing manual quality inspection and lowering QA workforce costs significantly.
  • Sole engineer responsible for camera selection, labelling, architecture design, and training — true end-to-end ownership from concept to ROI.
YOLO-V5OpenCVRoboflowDeep LearningMachine Vision
Published Writing

Thoughts on AI,
Data & Engineering

Sharing practical insights from real production AI/ML work — not just theory.

🔧
How I'm Using n8n to Automate the Boring Parts of Data Science
A practical guide to automating repetitive data science workflows — from data ingestion to notifications, freeing time for high-impact model work.
Read on Medium
🚀
From Data to Deployment: Building End-to-End ML Pipelines
A comprehensive walkthrough of building production-ready ML pipelines — from raw data all the way to deployed, monitored models in production.
Read on Medium
📊
Bridging the Gap Between Data Science & Business
How to translate complex ML/AI findings into clear, actionable business narratives that stakeholders actually understand and act on.
Read on Medium
View All Articles on Medium →
Technical Toolkit

Full-Stack
Data Science Skills

From raw data ingestion to boardroom dashboards — I own the entire pipeline.

Languages & Libraries
PythonSQLScikit-learnPandasNumPyLangChainLangGraphSpacyNLTK
AI / GenAI / Agentic
Agentic AIMulti-Agent SystemsLangGraphRAG PipelinesGPT-4LLaMAYOLO-V5XGBoostProphetMonte Carlo
Cloud & DevOps
AWS SageMakerAWS LambdaS3 / EC2Vertex AIDockerGitHub ActionsIguazioApache Kafka
Data & BI
BigQuerySnowflakeMongoDBOracleTableauAWS QuickSightAmazon AthenaDBeaver
Industry Domains
📱
Telecommunications
Vodafone · Customer behaviour & Finance analytics
🛒
Retail & E-Commerce
Home Depot · Office Depot · KPI, forecasting & CX
💰
Fintech & Capital Markets
MarketsMojo · Equity intelligence & portfolio analytics
🏭
Automotive & Manufacturing
Mercedes-Benz · Tata Technologies · Vision & IoT
Education
Post-Grad Diploma in Data Science
Simplilearn × Purdue University · 2017
Bachelor of Engineering
RTM Nagpur University · 2012
My Philosophy

"I don't build models to win Kaggle competitions. I build models that increase revenue, reduce costs, and give business leaders the confidence to make faster decisions."

— Chetan Mundhe, Lead Data Scientist
🌏
Based in India · Open to Global Roles
Holds a US B1/B2 visa. Available for senior IC, lead, and manager roles in CX analytics, AI/ML, and data science — on-site, hybrid, or remote.
🤝
Pre-Sales to Production
Experienced in client-facing requirement gathering, creating project quotes, and delivering POCs — not just back-office modelling.
👥
Team Leadership
Successfully led cross-functional teams of data engineers and analysts on high-impact projects, aligning technical execution with business strategy.
🔬
Depth + Breadth
From streaming ML pipelines and computer vision to Agentic AI and customer propensity models — rare versatility without sacrificing depth.
Open to Opportunities

Let's Build Something
That Moves the Needle.

Looking for a senior data scientist who takes ownership, speaks business language, and delivers models that generate real ROI — I'd love to connect.

Available for new roles · 2025