Most data scientists are already busy with feature engineering, model evaluation, and statistical thinking. With Generative AI, there is another dimension added – building systems that can do text generation, information retrieval, and take multiple steps rather than merely give you a prediction. The skill sets involved might have more similarities than expected, but the technologies used and techniques employed are novel.
The Executive Post Graduate Certificate in Generative AI & Agentic AI offered by IIT Kharagpur's Department of Computer Science and Engineering is an attempt to bridge this very gap. The course is 8 months long and is done entirely online.
Why This Program Fits a Data Scientist's Background
A data scientist's existing skill set maps onto this curriculum more directly than it might into a generic AI course, since much of the work is still about evaluation, iteration, and measuring model behaviour against a baseline.
Statistical knowledge is transferred into LLM evaluation and finetuning, but not in scoring accuracy
Experience with Python and APIs is necessary, but not something to be taught from the very beginning
Experience with feature engineering comes in handy while working on retrieval pipeline for RAG models
Experience with model evaluation is directly applicable to compare LoRA vs QLoRA finetuned models to the base one
Knowledge of structured data is the first step towards unstructured, multimodal data working
Curriculum: From Foundations to Deployment
The program follows a fixed progression, moving from core concepts to systems that run in production rather than staying at the prompting stage.
Foundations of GenAI and LLMs, including transformer architectures and foundation models
Advanced prompting and RAG techniques, including hybrid search and reranking
LLM fine-tuning and alignment through PEFT, LoRA, and QLoRA
Multimodal and agentic AI, including agents’ workflows involving planning and tool usage
Deployment, optimization and AI safety, including latency, cost, and reliability in production
Projects: What Gets Built, Not Just Taught
The course is organized into five deliverables instead of separate tasks, and thus, each module is designed such that it delivers an output which is functional.
Enterprise RAG System with an evaluation pipeline to assess the retrieval accuracy
Open-source LLM with tuning using LoRA/QLoRA and deployment as an API
Multi-Agent System to support multiple step task execution
Containerized and monitored Generative AI API to handle actual traffic
2-week industry capstone project with faculty guidance
Program Details at a Glance
Fee: ₹1,99,000 including taxes
Seat block: ₹10,000
EMI: starting from ₹6,825 per month
Qualifications for application: bachelor’s degree with minimum 50% marks
Others: eligible if you have at least 2 years of experience in technology domain
Skills needed: knowledge of writing functions in Python, data structures, working with APIs and reading tech documentations
Certificate awarded: on campus at the time of graduation at IIT Kharagpur, handed over by Programme Director & Institute management
Admission Process
Joining the program is easy and only requires four simple steps.
Enter your personal details and apply.
Be selected and get an offer letter.
Book your seat and submit all the necessary documents.
Make the remaining payment after verification.
Why Choose IIT Kharagpur for Generative AI
IIT Kharagpur brings a few things to this program that are hard to find combined in a single online course.
India's first IIT, founded in 1951, with 75 years of educational experience.
Ranked 5th in Engineering in NIRF 2025, amongst top-ranked universities in India
Professors who regularly publish papers in conferences like NeurIPS, ICML, ICCV, ACL, and IEEE, hence, the syllabus will be based on research and not tutorials
Classes conducted only by professors from the CSE department with no third-party instructors involved at any point in time
Certificate awarded to you after an on-campus ceremony and not just a digital one
Conclusion
The more challenging aspect for a data scientist in embracing generative AI is not the mathematical theory or the code but how to evaluate the different systems which have behavior distinct from the usual machine learning algorithms, i.e., where there is output variability, multiple-step agents and post-deployment monitoring is necessary.
This course has been designed with this in mind, by providing hands-on experience in developing systems like RAG pipelines, fine-tuned LLMs and multi-agent systems which can be deployed as opposed to isolated tasks. This will give the data scientist already well-versed in Python and modeling with statistics a structured entry into generative AI.