Build it. Fine-tune it. Ship it. A programme offered by the Continuing Education Programme (CEP), IIT Delhi.
Enquiry about programme
Six months to build real Generative AI systems — and open doors across AI engineering, research and product roles.
PROGRAMME OVERVIEW
This is not a tools workshop. You move from the mathematics under modern AI to the architectures defining its frontier — Transformers, fine-tuning, RLHF, RAG, multimodal AI and agentic systems — so you don’t just use Generative AI, you can build with it.
Industry vetted curriculum
10 modules across 4 phases, 86 hours. Every session is listed with exactly what it teaches. Expand any module to see the topics it covers — then head to Projects for the six shipped builds and the final capstone.
How is this module future-proof
Architectures change every year; the mathematics of how models learn does not. This is the foundation every future model rests on.
Module 1 · Foundations · 5 sessions
The mathematical language under modern AI.
How this module is helpful
Builds the linear algebra, probability and optimization you need to reason about how models learn.
Takes you through the neural-network training loop and backpropagation from first principles.
Grounds ML terminology — regression, classification, regularization, evaluation, K-Means, PCA — before the deep-learning modules.
Want to dive deeper into the details?
Ready to join the Certificate Programme in Generative AI (Batch-03) and take your first step towards success?
PROJECTS YOU WILL SHIP
Every module ends with a working artefact. Six graded builds accumulate into a portfolio-\ grade body of work - the kind that lets you walk into a GenAI interview and show, not tell. Projects shown are indicative and may be modified to fit the programme.
A fully connected network built in PyTorch or TensorFlow, trained on a task with its accuracy and loss visualised over epochs.
Ships: a trained neural network with accuracy and loss curves.
A pre-trained transformer fine-tuned on a text classification task, evaluated on a test set for accuracy, precision and recall.
Ships: a fine-tuned transformer with a precision/recall evaluation.
A simplified transformer built around self-attention and positional encoding, trained on a small dataset and compared against a pre-trained model.
Ships: a from-scratch transformer benchmarked against a pre-trained model.
PEFT applied to a large LLM for a specific task, with training time and resource usage compared against standard fine-tuning.
Ships: a PEFT fine-tune with a cost/resource comparison vs full fine-tuning.
A reward model trained on human-labelled data for text generation, with a reinforcement-learning loop that improves the LLM’s responses.
Ships: a reward model plus an RLHF loop that improves LLM outputs.
A comparison of open-source LLMs on reasoning tasks across zero-shot, few-shot, chain-of-thought and self-consistency prompting.
Ships: a benchmarked comparison of prompting strategies on reasoning tasks.
LEARNING OUTCOMES
Every concept is reinforced through a working build. By the end of the programme you have six shipped projects, a defended portfolio, and the hands-on confidence to take any GenAI idea from prototype to production.
01
Understand the mathematics — linear algebra, probability, optimisation and the neural-network training loop from first principles.
02
Master the architectures — Transformers, encoder-decoder stacks, attention and pretraining strategies: the modern LLM stack.
03
Fine-tune & align — SFT, PEFT, RLHF and reward modelling to shape LLMs to specific tasks and organisations.
04
Build RAG & agents — retrieval-augmented pipelines, tool use, planning, memory and multi-agent orchestration.
05
Go multimodal & efficient — vision-language models (CLIP, BLIP, SAM) and small, on-device model deployment.
06
Deploy responsibly — bias, safety, privacy and evaluation, to ship AI systems that hold up under audit.
Programme Coordinator
The programme is led by Programme Coordinator Prof. Tanmoy Chakraborty - an active author, editor and award-winner in NLP, LLMs and computational social science research.
Prof. Tanmoy Chakraborty
Yardi School of Artificial Intelligence & Department of Electrical Engineering Indian Institute of Technology Delhi Founder and lead of the Laboratory for Computational Social Systems (LCS2), a research group specializing in Natural Language Processing and Computational Social Science.
Prof. Tanmoy Chakraborty
Yardi School of Artificial Intelligence & Department of Electrical Engineering Indian Institute of Technology Delhi Founder and lead of the Laboratory for Computational Social Systems (LCS2), a research group specializing in Natural Language Processing and Computational Social Science.
TOOLS AND TECHNOLOGIES
You build in the tools the industry actually uses. Named tools are illustrative of the current industry stack; TRL, vector stores and evaluation frameworks are introduced in the relevant modules and may be substituted at faculty discretion. Tools shown are indicative and may be modified to fit the programme.
WHY NOW
Generative AI moved from research lab to enterprise core faster than any recent technology cycle. The gap between organisations that can build AI and those that can only buy it is widening — and the professionals who can build it are the scarcest resource in the market.
72%
Of employers report AI hiring difficulty, ahead of engineering & IT for the first time.
ManpowerGroup
ManpowerGroup Global Talent Shortage survey (39,000 employers, 41 countries), 2025
#1
AI skills are now the hardest technology skill to fill, for the first time in the annual survey.
ManpowerGroup
ManpowerGroup, most in-demand technology skills, 2025
3.2:1
Global demand-to-supply ratio for AI talent; LLM development is one of the most acute gaps.
Industry estimates
Industry AI-talent demand-to-supply estimates, 2026
34 - 41%
Projected compound growth of the Generative AI market through the early 2030s.
Analyst forecasts
Analyst forecasts (Grand View Research; MarketsandMarkets)
HOW YOU LEARN
A fully live online cadence — every concept reinforced through a working build, with an optional in-person immersion at IIT Delhi.
Live theory and tutorials across 4 phases — 86 hours led by IIT Delhi faculty and Programme Coordinator Prof. Tanmoy Chakraborty.
Six shipped projects and a final capstone: a client-ready GenAI/Agentic AI application, presented with demo, architecture, prompt flow and evaluation logic.
AI mock interviews with distinct interviewers for each round, plus an AI resume builder and review trained on GenAI hiring patterns — offered by Interviewbit (Varsity).
Admission is by the IIT Delhi Programme Coordinator based on eligibility and a screening review, so cohorts are matched in level. No prior AI experience required.
An optional in-person immersion at IIT Delhi, where the final capstone is presented live to faculty. Hybrid option available for those who cannot attend.
CAMPUS IMMERSION
This programme is delivered online, built to fit around a full-time job. An optional campus immersion adds an in-person experience at IIT Delhi, where the final capstone is presented live to faculty.
Indian Institute of Technology - Delhi
An optional in-person immersion for interaction between faculty and learners on the IIT Delhi campus. Note: Travel and accommodation cost will be borne by the learners. IIT Delhi will not be responsible for the same. A hybrid option will be open to candidates who are unable to come on campus.
In-person sessions and hands-on labs inside IIT Delhi's research facilities.
Interact directly with IIT Delhi faculty, researchers, and industry mentors.
Build a lasting network with a driven cohort of working professionals.
Certificate
On completion, participants receive a certificate from the Continuing Education Programme (CEP), IIT Delhi. Two certificate types are issued, based on attendance and assessment performance. Only e-certificates are issued for this programme.
Certificate of Successful Completion
Awarded to candidates who score at least 60% marks overall and have a minimum attendance of 80%.
Certificate of Participation
Awarded to candidates who score less than 60% marks overall and have a minimum attendance of 80%.
WHO SHOULD ENROL
Admission is selective. You will fit if you have a working foundation in engineering, computing or a technical discipline. No prior AI experience required — the curriculum builds from foundational mathematics upward.
6 Months
DURATION
Online live
FORMAT
86 hrs
LEARNING HOURS
17th Oct'26
STARTS
01
Core engineering & computing
02
B.Sc / BCA students
03
STEM graduates & professionals
Admissions
Follow the steps to understand how cohort information works.
Cohort Calendar
Admission Process
Starting at
₹10,137/month
One investment, a career-long return - an IIT Delhi credential and the applied GenAI skills to put your learning to work from day one.
Payment
Balance + GST as one-shot, or EMI via NBFC partners.
Refund Policy
Note on instalments
FAQ
Honest answers to the questions that might be holding you back.
Build real Generative AI systems — become the one top companies are hiring for.
