Skip to content
SDB

About me

Subhendu Datta Bhowmik, CSCP

Senior Director, IEP.AI Technology Office at Cognizant — 30 years spent carrying organisations from how the work has always been done to how it could be done.

SDB

Subhendu Datta Bhowmik

Senior Director, IEP.AI Technology Office

CurrentlyCognizant
In practice since1996
Based inKolkata, India
FocusAI · Platforms · Industry 4.0
StudiedM.Tech Data Science, BITS

The short version

I build the systems that businesses quietly depend on, and I have been doing it since before the web had a business model.

30 years in Information Technology has meant four near-complete reinventions of the toolkit — client–server and hand-built ERP, enterprise integration, cloud, and now model-led systems. What has not changed is the shape of the work: understand the domain deeply, choose the constraints deliberately, and leave behind something the next team can safely modify.

Most of that career has been spent at enterprise scale. I spent thirteen years as global delivery manager and chief architect on a metals manufacturer’s estate — roughly 500 specialists, four data centres, four continents — then six years directing an Oracle solution practice across communications, hi-tech, manufacturing, life sciences, healthcare and logistics as those estates moved to the cloud. Overseas delivery took me through the USA, Australia, Brazil, Ireland, Switzerland, Hungary, the UAE, Singapore and Tanzania. The interesting problems were rarely the algorithms. They were migrations that could not take the business offline, integrations across teams that had never met, and the long negotiation between what the architecture wants and what the calendar allows.

Today I am Senior Director, IEP.AI Technology Office at Cognizant, leading the strategy, architecture and delivery of enterprise-grade IP platforms across generative and agentic AI, Industry 4.0, digital twins, XR, conversational AI, robotics and responsible AI. That includes Cognizant Neuro® AI Enterprise Core — the lifecycle layer for context engineering, prompt orchestration, agent frameworks, knowledge-graph grounding, vector search and model governance — alongside AutoML platforms, supply-chain digital twins and enterprise copilots that plug into SAP, Oracle, Salesforce, ServiceNow and the major cloud hyperscalers.

The question I ask of all of it is the same one I have asked for three decades: will this still be a good idea in five years? This site is where I write down the answers — the decisions that held, the ones that did not, and what I would tell a younger engineer standing at the near end of the bridge.

Years in IT
30+Years in IT
Countries worked in
10Countries worked in
Largest team led
500Largest team led
Continents delivered across
4Continents delivered across

How I work

01

Boring where it counts

Novelty belongs at the edges of a system, never at its core. The database, the identity model and the deployment path should be the least surprising things you own.

02

Optimise for change, not for launch

Every architecture is judged years after go-live. The real metric is how cheap the tenth change was, not how fast the first release shipped.

03

Write the trade-off down

A decision without a recorded rationale becomes folklore within eighteen months, and folklore cannot be revisited when the constraints change.

04

The org chart ships in the product

Integration problems are usually team problems wearing a technical disguise. Fix the contract between the people and the interfaces follow.

Platforms & IP

What I build for enterprises now

Production-grade PaaS and SaaS intellectual property — generative and agentic AI, digital twins and copilots — designed to sit alongside the systems of record organisations already trust.

01Foundational AI lifecycle platform

Cognizant Neuro® AI Enterprise Core

The layer that makes large-scale generative and agentic AI adoption governable — from context and agents through grounding, search and responsible-AI controls — so enterprises can ship copilots and autonomous workflows that connect to real systems of record.

  • Context engineering & prompt orchestration
  • Agent frameworks
  • Ontology & knowledge-graph grounding
  • Vector search
  • Model governance & responsible AI
  • Enterprise system integration
02Conversational AI at enterprise depth

AI Virtual Assistant & Enterprise Copilots

End-to-end bot lifecycle management with proprietary NLP/NLU, visual bot building, training pipelines and on-premise or cloud deployment — plus multilingual, multimodal assistants with live-agent integration.

  • NLP / NLU lifecycle
  • Visual bot building & training
  • Multilingual & multimodal interfaces
  • Live agent handoff
  • On-prem or cloud deployment
  • Deep enterprise app connectivity
03From data to governed models

AutoML & Generative AI Platform

Automated data loading, profiling and visualisation through feature engineering, hyperparameter tuning, model interpretability and deep-learning capabilities — extended into LLM fine-tuning, multimodal AI and MLOps/LLMOps.

  • Automated data prep & profiling
  • Feature engineering & AutoML
  • Hyperparameter tuning
  • Model interpretability
  • LLM fine-tuning & multimodal AI
  • MLOps / LLMOps
04Physical operations, digitally twinable

Industry 4.0 Digital Twin & Control Towers

Real-time digital counterparts of physical objects and processes — IoT streaming, time-series analytics, ML predictions, GIS and XR visualisation, and robotics for smart manufacturing and logistics.

  • Real-time IoT streaming
  • Time-series analytics & ML prediction
  • GIS & XR visualisation
  • Supply-chain control towers
  • Smart manufacturing & logistics
  • Robotics integration

Capabilities

What I bring to a problem

Depth in AI and enterprise architecture, breadth across the platform, data and operations stack, and enough leadership experience to make either one land inside a real organisation.

Generative & agentic AI

  • Context engineering & prompt orchestration
  • Retrieval-augmented generation, hybrid search
  • Agent frameworks — LangChain, LangGraph, CrewAI
  • LLM fine-tuning & evaluation
  • Ontologies, knowledge graphs, semantic AI
  • Responsible AI & model governance
  • OpenAI GPT, Azure, Amazon Bedrock, Claude, Gemini

Data, ML & platforms

  • AutoML & feature engineering
  • Transformers, CNN, RNN, GAN
  • MLOps & LLMOps
  • Vector databases
  • Streaming & time-series analytics
  • AI-assisted engineering — Cursor, Claude Code, Copilot, Amazon Q, Gemini

Industry 4.0 & physical AI

  • IoT platforms & digital twins
  • Supply chain control towers
  • Robotics — ROS2, MuJoCo, SLAM
  • Robotic arm ML modelling, ACT policy
  • XR/AR/VR — Unity, Unreal
  • Generative media — Veo, Sora, Suno, Runway

Enterprise & cloud

  • Cloud architecture — AWS, Azure, GCP, OCI
  • Oracle E-Business Suite, OTM/GTM, IoT PaaS
  • Exadata, RAC, WebLogic, Autonomous DB
  • SAP, Salesforce, ServiceNow integration
  • Full-stack — Python, Java, JavaScript, APEX, ADF
  • Legacy modernisation & SOA / Internet of Things gateway patterns

Supply chain & operations

  • Order to cash, procure to pay, finance
  • Advanced supply chain planning
  • Enterprise asset management & field service
  • Transportation & global trade management (OTM/GTM)
  • Certified Supply Chain Professional (CSCP)
  • Manufacturing execution, RFID & barcode

Leadership

  • Technology strategy & industry solutions
  • IP, product & venture incubation
  • Global delivery at 500-person scale
  • Architecture governance & security compliance
  • Team building, mentoring & communities of practice
  • Sales, marketing & analyst enablement

The journey

1996 to today

Five roles, four frontiers, one continuous education — and the projects that taught each lesson.

  1. 1996 — 2001

    India · Tanzania

    ERP Solution Architect

    Tata Consultancy Services

    Building financial and billing systems when an ERP was still something you largely wrote yourself. Fixed assets, cash management and service-order billing for Tata companies and overseas telecom — the lesson that has outlasted every framework since: data outlives the code that produced it.

    Selected work

    • Fixed asset and cash management modules for AVALON ERP, delivered to Tata Tubes and Tata Bearings
    • Service order billing system for TTCL, Tanzania
    • Oracle Applications implementations for Ares Serono and Duck Back
    • Oracle Applications upgrade for GE Capital Aviation Services
    • Oracle Forms
    • Reports
    • PL/SQL
    • Unix
    • Workflow
  2. 2001 — 2002

    India · USA

    Oracle CoE Lead

    Cognizant

    A first turn towards the front of the business — incubating solutions, proving them in the room, and learning that architecture has to survive a commercial conversation before it ever reaches production.

    Selected work

    • Pre-sales, consulting, proof-of-concept and solution incubation across the Oracle Applications portfolio
    • Lead solution architect and designer for Centre of Excellence reference solutions
    • ERP fitment study for Case New Holland (CNH) in the USA, plus additional customer engagements
    • Oracle Apps
    • Solution design
    • Pre-sales
    • POC
  3. 2002 — 2015

    USA · Hungary · Australia · Brazil

    Global Delivery Manager & Chief Architect

    Tata Consultancy Services · Alcoa

    Thirteen years running global delivery for a metals manufacturer: around 500 specialists, four data centres and four continents on one release calendar. Concurrent work as OTM and EAM solution architect for energy, utilities and mining clients taught that integration is rarely only a technical problem.

    Selected work

    • Alcoa — application support and enhancement across four data centres (North America, Europe, South America, Australia); Best Team awards from the client CIO for three consecutive years
    • TIBCO to Oracle Fusion Middleware migration, and Interconnect to OSB migration
    • Exadata platform migration, data-centre consolidations, GDW & Hyperion upgrades, and the RPG divestiture programme
    • Schlumberger — end-to-end OTM 6.2 (order base, shipment planning, contract & rate management, GTM) with a custom financial suite for SAP and Lawson, plus contract/rate loaders, offline transport-request tools and custom LDAP login pages
    • City West Water, Australia — Oracle EAM and Field Service integrated with GIS for linear asset lifecycle management
    • Vale, Brazil — ERP stabilisation and technology roadmap; system performance improved ~38% and availability by ~10%
    • Kirloskar Oil Engines — six Oracle Applications instances consolidated into one, followed by application upgrade
    • OTM/GTM
    • Fusion Middleware
    • SOA & BPEL
    • Exadata
    • Hyperion
    • OBIEE
    • RFID & MES
  4. 2015 — 2021

    Global portfolio · Communications, Hi-tech, Manufacturing, Life Sciences, Healthcare, Logistics

    Oracle Solution Practice Director

    Cognizant

    Custody of an infrastructure and technology delivery portfolio spanning communications and media, hi-tech, manufacturing, life sciences, healthcare and logistics — while the same estate began its move to cloud, IoT, chatbots and predictive analytics.

    Selected work

    • Platform migrations, RAC and Exadata programmes, instance consolidation, capacity management and performance tuning
    • E-Business Suite and OTM install, upgrade, cloning, patching and disaster-recovery setup; Identity management, SSO and SSL; WebLogic administration
    • Autonomous Database adoption and 19c upgrades across the portfolio
    • Oracle digital transformation — IoT, blockchain, machine learning, Oracle Chatbots, cognitive and predictive analytics for E-Business
    • On-premises to Oracle Compute Cloud migrations with mobile and web front ends for Oracle ERP
    • Hands-on depth across EBS Order to Cash, Procure to Pay, Finance, EAM, Advanced Supply Chain, Field Service and ECC (5–6 large implementations; 10+ upgrades as lead), plus 3–4 large OTM/GTM 6.x programmes
    • Application development across full-stack JavaScript, Python, Java, APEX and ADF, alongside Forms, Reports, Workflow, BI Publisher, OBIEE and Endeca
    • Oracle EBS 11i–12.2.9
    • OTM/GTM 6.x
    • OCI
    • Exadata
    • RAC
    • WebLogic
    • Oracle IoT
    • Chatbots
  5. 2021 — today

    Enterprise IP platforms · Generative & Agentic AI · Industry 4.0

    Senior Director, IEP.AI Technology Office

    Cognizant

    Strategy, architecture and delivery for enterprise-grade AI and digital-engineering IP — positioning ventures with sales and marketing, mentoring innovators, and building production platforms that enterprises can actually govern. Thirty years of fundamentals turned out to be exactly the right preparation.

    Selected work

    • Cognizant Neuro® AI Enterprise Core — end-to-end AI lifecycle management: context engineering, prompt orchestration, agent frameworks, ontology and knowledge-graph grounding, vector search, model governance and responsible AI controls
    • Technology strategy and industry solutions across digital engineering, conversational AI, ML/DL, blockchain, AR, cognitive automation, metaverse, IoT and digital twins
    • AutoML and generative AI platforms for feature engineering, model training, LLM fine-tuning, multimodal AI and MLOps/LLMOps
    • Industry 4.0 supply-chain digital twins and control towers — real-time IoT streaming, time-series analytics, ML prediction, GIS and XR visualisation, robotics for smart manufacturing and logistics
    • Enterprise copilots and multilingual, multimodal virtual assistants with live-agent handoff and deep connectivity to SAP, Oracle, Salesforce, ServiceNow and major cloud hyperscalers
    • Market differentiation, technology enablement across the business unit, R&D roadmaps, budgets and regulatory/security adherence for PaaS and SaaS IP
    • Generative AI
    • Agentic AI
    • LangGraph
    • Knowledge graphs
    • Digital twins
    • XR
    • ROS2

Industries

Where the systems ran

Domain knowledge is the part of this work that does not transfer from a conference talk. Each of these took a programme to learn properly.

  • Manufacturing
  • Mining & metals
  • Logistics & supply chain
  • Utilities
  • Communications & media
  • Hi-tech
  • Life sciences
  • Healthcare

Geographies

Where the work took me

Delivery across four continents, on assignment in ten countries. Time zones turn out to be an architectural constraint like any other.

  • India
  • USA
  • Australia
  • Brazil
  • Ireland
  • Switzerland
  • Hungary
  • UAE
  • Singapore
  • Tanzania

Credentials

Studied, certified, recognised

A production engineer by training, a data scientist by retraining, and a supply chain professional by certification — which explains most of what I find interesting.

Education

  • Robotics — Executive Program

    IIT Delhi · 2026

  • M.Tech, Data Science

    BITS Pilani · 2021 — 2023

  • B.E., Production Engineering

    Jadavpur University · 1992 — 1996

Certifications

  • OpenAI Consultative Solution Practitioner

    OpenAI · Sep 2026

  • ChatGPT Deployment Practitioner

    OpenAI · Sep 2026

  • ChatGPT Solution Practitioner

    OpenAI · Sep 2026

  • Codex Deployment Practitioner

    OpenAI · Sep 2026

  • Codex Solution Practitioner

    OpenAI · Sep 2026

  • Gemini Enterprise Agent Development

    Google — Certified Partner Specialist · Sep 2026

  • CSCP — Certified Supply Chain Professional

    ASCM, USA · 2009

  • Oracle Compute Cloud Associate

    Oracle

  • Oracle Autonomous Database Specialist

    Oracle

  • OCI Development Associate

    Oracle

  • Internet of Things

    Professional certification

  • Machine Learning

    IBM & Stanford University

Recognition

  • Best Team award from the client CIO, three consecutive years, on the Alcoa global delivery programme
  • Cognizant Vibe Coding Event Champion — AI-assisted engineering with Cursor, Claude Code, GitHub Copilot, Amazon Q and Gemini
  • Google Skills for Partners — Diamond League (6,633+ points), with skill badges across ADK, Gemini Enterprise Agent Platform, Vertex AI and Responsible AI

Let’s talk

Thirty years of notes, and I am still most interested in the next problem.