◆ SUBHODEEP MITRA
Principal AI Architect · 11+ years

Designing enterprise-grade GenAI & Multi-Agentic AI systemsthat ship fast, govern responsibly & scale globally.

I design and lead production grade  AI solutions on Microsoft Azure, Google Gemini Enterprise & GCP, AWS, Databricks and Snowflake — multi-agent LLM systems, RAG and Responsible AI — translating ambiguous business problems into reliable, cost-aware architectures.

Explore the work LinkedIn Download Resume
11+
Years
30–40
Engineers led
5
Cloud platforms
Subhodeep Mitra — Principal AI Architect
AZURE · AWS · DATABRICKS · GCP · SNOWFLAKE
GenAI · Agentic · RAG
◆ Azure OpenAI◆ GPT-4◆ LangGraph◆ AutoGen◆ Semantic Kernel◆ Bedrock Agents◆ Mosaic AI◆ MLflow◆ Pinecone◆ Weaviate◆ RAG◆ Multi-Agent◆ Responsible AI◆ Prompt Flow◆ Databricks◆ Azure OpenAI◆ GPT-4◆ LangGraph◆ AutoGen◆ Semantic Kernel◆ Bedrock Agents◆ Mosaic AI◆ MLflow◆ Pinecone◆ Weaviate◆ RAG◆ Multi-Agent◆ Responsible AI◆ Prompt Flow◆ Databricks
Selected outcomes

What production AI looks like in practice.

01

Azure-native GenAI delivery

Deployed OpenAI models via Azure AI Studio & Azure AI Foundry for multilingual text analysis and dynamic content generation with prompt engineering and evaluation workflows.

02

Enterprise GenAI on GPT-4

Architected production Generative AI automating product content and metadata generation at enterprise scale for a global FMCG leader.

03

Reusable RAG & agentic assets

Built Databricks-based RAG and multi-agent platform assets that accelerated enterprise AI adoption across teams.

04

Multi-cloud solution design

Designed and deployed multi-agent LLM workflows across Azure and AWS with built-in evaluation, scalability and reliability.

Trajectory

A decade of applied AI, end-to-end.

Focus:Data ScienceMachine LearningDeep LearningData ModellingPredictive AnalyticsNLP & Text AnalyticsComputer VisionStatistical ModellingTime-Series ForecastingKnowledge Graphs
Years2026Tiger2025Tiger2024Tiger2023Cognizant2022Capgemini2021Capgemini2020Capgemini2019Capgemini2018Capgemini2017Capgemini2016Capgemini2015Capgemini
Aug 2024 — Present

Lead AI/ML Engineer — GenAI, Agentic AI & Data Science

Tiger Analytics · Pune / Chennai, India
GenAIAgentic AIRAGLangGraphDatabricksAzureAWSPythonPyTorchTensorFlowspaCy
Impact
15–20
Data scientists led across GenAI & NLP delivery
Impact
2
Hyperscalers in production — Azure + AWS multi-agent stacks
Impact
↑ TTM
Reusable Databricks GenAI/RAG assets cut enterprise time-to-market
  • Designed and deployed multi-agent AI systems across Azure and AWS — RAG, planner-executor and tool-calling patterns wired to enterprise data with evaluation, security and reliability baked in.
  • Lead and mentor a cross-functional team of 15–20 Senior Data Scientists and Data Scientists; own GenAI & NLP solution delivery end-to-end.
  • Built reusable Databricks GenAI / RAG and Agentic AI assets that standardize patterns and accelerate enterprise AI time-to-market.
  • Design & develop NLP models for text classification, custom entity recognition, relationship extraction, summarization, topic modeling, semantic search and reasoning over Knowledge Graphs using spaCy, TensorFlow and PyTorch.
  • Build image-recognition and video-analysis models with state-of-the-art deep learning and OpenCV; apply Generative AI to a wide range of business problems.
Feb 2023 — Aug 2024

Senior Data Scientist — AI & Analytics

Cognizant Technology Solutions · Kolkata, India
Deep LearningNLPComputer VisionKnowledge GraphsTensorFlowPyTorchspaCyOpenCVGenAI
Impact
4
Modalities shipped — text, speech, image, video
Impact
+Accuracy
Lift on customer models through technique-fit redesigns
Impact
E2E
From data dictionary to deployed model with business impact framing
  • Hands-on problem solver with a consultative approach — applied AI, ML and Deep Learning to business challenges, finding the combination of techniques that best fit the problem and lifted model accuracy for greater business impact.
  • Partnered with domain and customer teams to translate business context and data dictionaries into deep learning solutions, and estimated downstream business impact of model deployment.
  • Designed, developed and deployed Deep Learning models on TensorFlow and PyTorch across text, speech, image and video data.
  • Built NLP pipelines for text classification, custom NER, relationship extraction, summarization, topic modeling, semantic search and reasoning over Knowledge Graphs using spaCy, TensorFlow and PyTorch.
  • Developed image-recognition and video-analysis models with OpenCV and SOTA deep learning algorithms; deployed Generative AI solutions across diverse business problems.
Jul 2020 — Feb 2023

Senior Consultant (Technical) — AI/ML, NLP & Group Sales (TLC)

Capgemini Technology Services · Kolkata, India
NLPAPEXBig Deals RadarAI ChatbotPythonFlaskREST APIsTensorFlowPostgreSQLNode.js
Impact
2
Flagship Group-Sales platforms owned — APEX & Big Deals Radar
Impact
Prod
Led deployment of Big Deals Radar to production servers
Impact
Global
Direct line into CG Global Sales Strategy leadership
  • Implemented NLP and text-recognition for account planning and business-decision automation — smart keyword extraction, web scraping with Node.js, PostgreSQL data layer, and ongoing sentiment analysis.
  • Built the AI Chatbot for the Big Deals Radar platform and partnered with the CG Global Sales Strategy team to convert business context into data analytics and ML solutions.
  • Owned the Account Planning & Execution Tool (APEX) — automation, data visualization and analysis across sales accounts to strengthen sales strategy.
  • Stack: Python, Flask, REST APIs, TensorFlow, image recognition and big-data modelling; led deployment of Big Deals Radar applications to production servers.
Oct 2018 — Jul 2020

Team Leader / Associate Consultant — Burberry Group PLC

Capgemini · Kolkata, India
SAP BODSSAP BI/BWPredictive AnalyticsMLDeep LearningAndroidPythonTensorFlowImage Recognition
Impact
6
Engineers led as Team Leader on the Burberry account
Impact
2 yrs
Primary liaison between app dev and global business stakeholders
Impact
1
Fashion Finder Android app — DL-powered catalogue match
  • Managed a 6-person team as team leader on the Burberry Group PLC account — primary liaison between app dev and global business stakeholders.
  • Built supply-chain allocation & replenishment and sales-forecasting models using predictive analytics, machine learning and data-mining techniques to forecast seasonal sales.
  • Delivered SAP BODS & Business Intelligence (BI/BW) job creation and testing across the data pipeline.
  • Developed the 'Fashion Finder' mobile app (Android) using Deep Learning and AI — recognizes a fashion article and suggests matching merchandise from the client catalogue.
  • Stack: Python, TensorFlow, image recognition, big-data modelling, SAP BI/BODS/BW.
Oct 2017 — Oct 2018

Associate Consultant — Control-M Batch Scheduling

Capgemini · Kolkata, India
Control-MWLSTShell ScriptingUnixLinuxMainframeAS400ServiceNowMS VisioData Analytics
Impact
3,000+
Control-M jobs documented and maintained in production
Impact
Multi-OS
Unix, Linux, Intel, Mainframe and AS400 schedules owned
Impact
Trainer
Delivered Control-M training to app & dev teams
  • Administered 3,000+ Control-M jobs across Production, QA and Development on Unix, Linux, Intel, Mainframe and AS400 estates — ensuring 99.9% batch schedule reliability.
  • Authored automated WLST and shell scripts for deployment, monitoring, health checks and domain creation — reducing manual intervention by 40%.
  • Analysed batch runtimes and failure patterns to build predictive workload models that cut SLA breaches by 25% and improved scheduling efficiency.
  • Built real-time dashboards and reports from Control-M EM databases to track throughput, resource trends and capacity — enabling data-driven operations decisions.
  • Delivered Control-M training to app and dev teams; administered security policies for Operations and Production Control staff.
  • Maintained end-to-end documentation, MS Visio flow diagrams and ServiceNow request workflows for 3,000+ production jobs.
Oct 2016 — Oct 2017

Senior Software Engineer — Oracle & SQL Server DBA

Capgemini · Kolkata, India
Oracle 11g/12cSQL Server 2016RMANData GuardVERITAS NetBackupSQL*LoaderData PumpShell ScriptingData Modelling
Impact
11g/12c
Oracle & SQL Server installations, migrations and capacity planning
Impact
RMAN
Hot/cold backup, cloning and disaster recovery via Data Guard
Impact
Tuning
Performance tuning across apps, memory, disk and OS
  • Administered Oracle 10g/11g/12c and SQL Server 2016 across Unix and Windows — handling installations, migrations, capacity planning and automated backup strategies.
  • Performance-tuned production databases across application, memory, disk and OS layers — resolving bottlenecks and cutting average query latency by 30%.
  • Designed relational data models and normalised schemas for reporting and analytics workloads; built complex SQL and stored procedures powering operational dashboards.
  • Built statistical trend models from AWR and SQL Server DMVs to forecast table growth, index bloat and I/O bottlenecks — shifting capacity planning from reactive to proactive.
  • Deployed Physical and Logical standby databases via RMAN and Data Guard with shell automation; executed hot/cold backup, recovery and cloning via VERITAS NetBackup.
  • Owned high-availability cluster administration, incident and change management, security compliance, and weekly DB service reporting to the client.
Aug 2015 — Sep 2016

Software Engineer — Cloud RIM & Operations

Capgemini · Kolkata, India
Cloud RIMITILServer MaintenanceIncident ManagementSecurity & ComplianceExcel Analytics
Impact
ITIL
Foundation-level certified; process-first delivery
Impact
24x7
Cloud Remote Infrastructure Management & operations monitoring
Impact
Client+
Received formal client appreciation for delivery
  • Delivered cloud-based Remote Infrastructure Management (RIM) and 24×7 operations monitoring on the ITIL framework — maintaining enterprise SLA targets.
  • Handled error resolution, server maintenance and ticket triage across the managed cloud estate — reducing average incident resolution time by 15%.
  • Created Excel analytics and pivot models tracking incident frequency, MTTR and uptime trends — surfacing insights that strengthened client SLA reviews.
  • Correlated CPU, memory and storage utilisation metrics to identify seasonal load patterns and right-size provisioning — optimising cloud resource spend.
  • Drove security and compliance management across internal audit cycles — ensuring zero critical findings during review periods.
  • Authored work instructions and knowledge-transfer documentation; earned ITIL Foundation certification and formal client appreciation for delivery excellence.
Capabilities

The stack behind the architecture.

Agentic AI

Multi-agent orchestration, planner-executor patterns, tool-calling. LangGraph, AutoGen, Semantic Kernel — across Azure, AWS and Databricks.

Generative AI

LLMs, RAG, vector embeddings, semantic search, prompt engineering, evals, guardrails and Responsible AI.

Azure AI

Azure AI Studio, Azure OpenAI (GPT-4), Azure ML, Cognitive Services, AI Search, Prompt Flow.

AWS & Databricks

Bedrock (Claude, Llama, Titan), Bedrock Agents, SageMaker. Mosaic AI Agent Framework, Vector Search, MLflow, Unity Catalog.

AI Security & Governance

Responsible AI, prompt-injection defense, PII redaction, Content Safety, RBAC, audit trails — EU AI Act / NIST AI RMF aligned.

MLOps & LLMOps

CI/CD, Docker, Kubernetes, LangSmith, AI Foundry tracing, eval harnesses, drift monitoring, cost telemetry.

Data & Knowledge

Vector DBs (Azure AI Search, Pinecone, Weaviate), knowledge graphs, hybrid retrieval, Delta Lake, Lakehouse design.

ML & NLP

Classification, NER, summarization, topic modeling, computer vision and predictive modeling.

On the record

Demo-Deep Learning Based Fashion Finder Mobile Application back in 2017 .

2007 — 2015 · Academic & Research Experience Background

Where the curiosity first took shape.

Before enterprise GenAI, there were soldering irons, turbo-alternators and grid graphs. A short archive of early projects, papers and research — the foundational chapters that led from electrical engineering in India to systems engineering research at Boston University.

Intelligent Home Automation System (IHAS)
◆ IIT Kharagpur · Kshitij ’09 · EurekaAug 2008 — Jan 2009
Aug 2008 — Jan 2009
IIT Kharagpur · Kshitij ’09 · Eureka

Intelligent Home Automation System (IHAS)

Co-authored and presented a paper proposing an intelligent domestic power system tightly coupled with Smart Metering and an Intelligent Grid — premise-level metering, producer-consumer coordination and demand-side energy management.

  • Selected in the top 13 papers across all engineering departments at India's largest techno-management festival
  • Awarded Certificate of Appreciation, Eureka Paper Presentation Competition
  • Framed four production levers: energy management, minimum power wastage, grid coordination, cheaper power
Top 13 · IIT Kharagpur
Digital Heartbeat Counter
◆ Bio-electronics · EE-893 · Team LeadOct 2008 — Mar 2009
Oct 2008 — Mar 2009
Bio-electronics · EE-893 · Team Lead

Digital Heartbeat Counter

Led a 3–5 week bio-electronic hardware build that measured pulse rate from the wrist using a hand-fabricated piezo-buzzer sensor — circuit design, PCB assembly, sensor calibration and end-to-end validation.

  • Team lead — owned circuit design, sensor fabrication and bench testing
  • Delivered an economically feasible bio-electronic alternative to commercial pulse monitors
  • Final outcome: working prototype, accurate beat-counting, minor sensor refinement noted as next step
Spring 2009 · EE-893 Project
From the archive
With the EE-893 team and the project professor — demonstrating the heartbeat counter live on a peer's wrist, Spring 2009.
With the EE-893 team and the project professor — demonstrating the heartbeat counter live on a peer's wrist, Spring 2009.
Graduated B.P. Poddar Institute · GPA 3.6 / 4.00
◆ WBUT · B.Tech, Electrical EngineeringAug 2009
Aug 2009
WBUT · B.Tech, Electrical Engineering

Graduated B.P. Poddar Institute · GPA 3.6 / 4.00

Completed Bachelor of Technology in Electrical Engineering under West Bengal University of Technology — capping the bio-electronics, power-systems and IHAS work above into a single undergraduate arc.

  • GPA: 3.6 / 4.00 across the full four-year electrical engineering curriculum
  • Prior industrial training at BHEL (Jul–Aug 2007) on Indo-German Siemens 250 MW turbo-alternators — Excellent Certificate, ref. C-07/357
  • Proficient in C; working knowledge of MATLAB R2009 and Java
Degree awarded · Aug 2009
Graduate research, teaching & adaptive control
◆ Boston University · M.S. Systems EngineeringSep 2009 — May 2011
Sep 2009 — May 2011
Boston University · M.S. Systems Engineering

Graduate research, teaching & adaptive control

Two academic years at Boston University on a Systems Engineering MS track (GPA 3.1 / 4.0 through Fall ’09). Original target graduation May 2011; the program was ultimately left incomplete after the research phase.

  • Fall 2009 — Courses:   Probability & Stochastic Processes; Dynamic Systems Theory; Linear & Non-linear Optimization.   Electronics Lab Assistant: installed lab instrumentation, supported faculty and students; commended for flexibility and work excellence
  • Spring 2010 — Courses: Sustainable Power Systems; Dynamic Programming; Discrete Event & Hybrid Systems; Non Linear Systems & Control;  Grader, Electronics Circuits: evaluated coursework end-to-end
  • Summer 2010 — Research under Prof. Hua Wang on 'Complex Network Representation of Power System Networks', merging mathematical physics with grid analysis
  • Spring 2011 — Reviewed and gave feedback on the Adaptive Control Toolbox (ACT) for Mathpros (Dave Wakstein) under Prof. Sean Andersson
MS program · incomplete
From the archive
With peers in the Boston University Intelligent Mechatronics Laboratory, interacting with the lab's mobile robot 'THREEPIO'.
With peers in the Boston University Intelligent Mechatronics Laboratory, interacting with the lab's mobile robot 'THREEPIO'.
PG Diploma in Industrial Automation
◆ NIELIT Calicut · Govt. of India2012 — 2013
2012 — 2013
NIELIT Calicut · Govt. of India

PG Diploma in Industrial Automation

Pivoted back from research into hands-on industrial systems — earned a one-year Post-Graduate Diploma in Industrial Automation from NIELIT (Govt. of India), bridging the Boston University control-systems work into PLC / SCADA practice.

  • Govt. of India accredited PG Diploma — industrial control, PLC, SCADA and instrumentation
  • Set the stage for the move into enterprise IT and, eventually, AI/ML
Govt. of India · accredited
From AI, Robotics & Control-systems engineer to enterprise IT
◆ Career transition · India2013 — 2015
2013 — 2015
Career transition · India

From AI, Robotics & Control-systems engineer to enterprise IT

The bridge years: moved from electrical & automation engineering into enterprise IT and data — the foundation for the next decade in AI/ML across Capgemini, Cognizant and Tiger Analytics.

  • Operationalised the academic control-theory and complex-network background into real client systems
  • Built the early software, data and architecture muscle that the post-2018 ML and GenAI work was built on
Bridge to AI/ML

Education

M.Tech, Data Science & Engineering
BITS Pilani · 2021–2023
PG Diploma, Industrial Automation
NIELIT, Calicut · 2012–2013
M.S., Systems Engineering (incomplete)
Boston University, USA · 2009–2012
B.Tech, Electrical Engineering
B.P. Poddar Institute, WBUT · 2005–2009

Certifications

Databricks Certified GenAI Engineer Associate
Associate level
Adv. Cert. Quantum Computing & AI/ML
IIT Roorkee · exp. Jul 2026
ITIL Foundation
Service management

Languages

EnglishHindiBengaliSpanishItalian

Publication

Intelligent Home Automation System — peer-reviewed research on scalable, AI-driven automation frameworks.

Let's build

Have an AI initiative in mind?

From a single PoC to a governed multi-agent platform across your cloud — happy to jam on architecture, evaluation, or team setup.

aideepkolkata@gmail.com linkedin.com/in/subhodeep-mitra-4ab30917
Kolkata · Hyderabad · Bengaluru, India

Protected against spam. No third-party trackers.

© 2026 Subhodeep Mitra
◆ Principal AI Architect