Hello๐Ÿ‘‹ I am

Koushik Deb โ€” AI Engineer, Researcher & Author

|

I'm an AI engineer at Accenture, building Generative-AI and Agentic-AI systems, and a PhD researcher at IIIT Kalyani studying how large language models behave across English, Hindi & Bengali. My work spans NLP, LLM evaluation and multilingual AI, and I'm the author of Keep It Simple: AI for Teachers.

13+Years in industry
12Publications
54+Citations
3Patents granted
3Best paper awards
What I'm working on now

Current Focus in AI & Research

Generative AI

Designing and shipping GenAI solutions for healthcare clients at Accenture, from the first use-case proposal through proof of concept to a deployed product.

Agentic AI

Systems that pair a language model with tools and retrieval (RAG), so answers come from a client's own documents and the work can run in production.

Large Language Models

Measuring how commercial LLMs respond to biased framing in a prompt. The latest study covered six models and 683,616 responses.

Multilingual AI & LLM Bias

Doctoral research at IIIT Kalyani into whether a model's fairness survives translation, tested in English, Hindi and Bengali.

Natural Language Processing

Sentiment analysis, topic modelling and threat detection in social-media text, plus a granted patent for running several NLP tasks offline on one device.

AI Evaluation

Deciding whether a model is fit for a particular job before anyone relies on it, with repeatable tests in research and practical checks at work.

๐Ÿ“˜

The book

Keep It Simple: AI for Teachers

A practical guide helping teachers and education professionals understand and use AI โ€” which tool to reach for, what to ask it, and what to check before anything reaches students. Published September 2026.

Explore the book โ†’
01 ยท Who I am

About Koushik Deb

Koushik Deb โ€” professional portrait
๐Ÿง  GenAI & Agentic AI
๐ŸŒ EN ยท HI ยท BN

I wear two hats. By profession I am an AI Engineer and Tech Lead at Accenture, where I design and ship Generative-AI and Agentic-AI solutions for global healthcare clients โ€” proposing new use-cases, building POCs and turning them into deployed products.

In parallel, I am a part-time PhD candidate at IIIT Kalyani (final year), researching bias detection and mitigation in multilingual large language models across English, Hindi and Bengali โ€” how prompts manipulate models, and how to measure and reduce that vulnerability.

Across 13+ years at Accenture, Cognizant, IBM and Capgemini I have built software for Apple, IKEA, Cigna, Roche, Tiffany & Co. and Sanofi. Along the way I picked up three granted patents โ€” from offline NLP on devices to drone-augmented pest control โ€” and a habit of tinkering with ESP32 boards and smart-agriculture hardware on weekends.

  • ๐ŸŽ“ Executive M.Tech, IIT (ISM) Dhanbad โ€” CGPA 9.03
  • ๐Ÿ”ฌ Research interests: LLM bias, NLP, sentiment analysis, topic modelling
  • ๐Ÿ“ Kolkata, West Bengal, India
02 ยท Where I've worked

Professional Experience

Accenture

Nov 2023 โ€” Present

AI Engineer ยท Tech Lead, GenAI

Generative-AI and Agentic-AI delivery for Cigna Healthcare and Roche, from proposal to deployed product.

Cognizant

Jan 2022 โ€” Nov 2023

Senior Developer โ€” ML/NLP

Machine-learning and NLP solutions for Apple Inc. and Inter IKEA Systems on Databricks.

Institute of Engineering & Management

Jul 2018 โ€” Jan 2022

Assistant Professor ยท Computer Science

Taught and researched NLP, sentiment analysis, topic modelling and information retrieval in Kolkata.

IBM India

Feb 2017 โ€” Jun 2018

Developer โ€” ML/NLP

Built machine-learning and NLP solutions for Tiffany & Co.

Capgemini

Jun 2013 โ€” Jan 2017

Software Engineer

Full-stack .NET delivery for Sanofi Aventis and a smart-energy billing platform for Sweden.

03 ยท What I study

AI Research

My AI research focuses on multilingual large language models, LLM bias, NLP and AI evaluation. Earlier work covered sentiment analysis, social-media threat detection and anomaly detection.

๐ŸŽ“

PhD Candidate ยท IIIT Kalyani since Aug 2022

Bias detection & mitigation in multilingual LLMs โ€” English เคนเคฟเคจเฅเคฆเฅ€ เฆฌเฆพเฆ‚เฆฒเฆพ โ€” final year, part-time alongside full-time industry work. Supervised by Dr. Imon Mukherjee (IIIT Kalyani), co-supervised by Dr. Debarshi Kumar Sanyal (IACS, Kolkata).

Latest ยท 2026

How prompts manipulate models: measuring cognitive bias vulnerability in commercial LLMs across languages

K. Deb, I. Mukherjee, D. K. Sanyal โ€” Innovations in Systems and Software Engineering (a NASA journal, Springer)

Measures how susceptible commercial LLMs are to cognitive-bias framing in prompts, and how that vulnerability shifts across languages.

04 ยท What I've published

Publications

12 publications 54+ citations 5 h-index 3 best paper awards
Google Scholar profile โ†—
2019

Predicting Radicalization & Civil-Unrest Threats from WhatsApp Groups

K. Deb, S. Paul, K. Das ยท Emerging Technology in Modelling and Graphics (Springer)

12 citations
2018

A Relative Kernel-Density Based Outlier Detection Algorithm

A. Wahid, A. C. S. Rao, K. Deb ยท SKIMA 2018 ยท ๐Ÿ† Best Paper Award

10 citations
2022

Crime by Crowd is Not Crime any More: Predictive Analysis of Riot & Radicalization Text

K. Deb et al. ยท Applications of Machine Intelligence in Engineering ยท ๐Ÿ† Best Paper, GCAIA

05 ยท What I invented

Patents & Innovations

โœ… Granted
๐Ÿ“ฑ

Multiple Natural Language Processing in Offline Mode

A single device doing real-time summarisation, topic detection and query-driven retrieval over text, audio, video and images โ€” fully offline.

Application 201931009037
โœ… Granted
๐Ÿ”ฅ

Fire Detection & Efficient Evacuation System

An IoT sensor network that detects fire, texts dynamically updated escape routes to occupants, and deploys firefighting robots on each floor.

Application 201931053496
โœ… Granted
๐Ÿš

Drone-Augmented Pest Control System

A camera-equipped drone swarm that spots crop pests with deep learning and neutralises them with targeted electromagnetic pulses โ€” no chemical spray.

Application 202431013004
06 ยท What I'm writing

Books

๐Ÿ“š Out now on Amazon ยท Kindle & Paperback

Keep It Simple: AI for Teachers

Which AI tool to use, what to ask, and what to check before it reaches students.

My first book โ€” a plain-language guide for teachers who keep hearing about AI and want to actually use it. No jargon, no hype, no computer-science degree required. Book one of the Keep It Simple series, with more on the way.

Read the first 19 pages free, buy it on Amazon, and leave your feedback โ€” I read every message. Open the book page โ†’

07 ยท What I use

Skills & Technologies

๐Ÿค– AI / ML

Generative AIAgentic AIRAGNLPMachine LearningDeep LearningData MiningLLM Evaluation

๐Ÿ’ป Languages

PythonC#JavaScriptDart

โ˜๏ธ Cloud & Platforms

AWSAzure AIDatabricksSQL Server

๐Ÿ› ๏ธ Tools

VS CodeCursorDevinPyCharmJupyterGit

๐Ÿ”Œ Hardware / IoT

ESP32RDK BoardsSmart AgricultureDrones
08 ยท Where I learned

Education & Certifications

Aug 2022 โ€” present

PhD, Computer Science & Engineering

IIIT Kalyani ยท Multilingual LLM bias detection & mitigation ยท final year (part-time) ยท Supervisor: Dr. Imon Mukherjee ยท Co-supervisor: Dr. Debarshi Kumar Sanyal (IACS)

May 2018

Executive M.Tech, Computer Science

IIT (ISM) Dhanbad ยท CGPA 9.03

2013

B.Tech, Information Technology

Govt. College of Engg. & Textile Technology, Serampore (WBUT) ยท CGPA 8.43

  • AWS AWS Certified AI Practitioner (AIF-C01)
  • Azure Azure AI Fundamentals (AI-900)
  • ML Machine Learning โ€” Thompson Rivers University (IEM-UEM Intl. Program)
  • IBM IBM Bluemix Certification
  • ITIL ITIL Foundation โ€” PEOPLECERT
  • Python Python Developer Certification โ€” VSkill
  • + Multiple Coursera AI/ML certifications
09 ยท Say hello

Connect With Koushik Deb

Whether it's a research collaboration, a GenAI problem, or just a good conversation โ€” my inbox is open. Write to me at:

๐Ÿ“ฎ koushikdeb88@gmail.com

Looking for something specific? Read about my AI research, browse publications and patents, see my books, or invite me to speak at a school, college or conference.