What This Workshop Is About
Generative AI is not just a trend — it is the current frontier of software engineering. Every company building products today is integrating LLMs, image generation, and AI-assisted workflows. This workshop teaches students to work with these tools the way industry practitioners do, not the way textbooks describe them.
Students leave with a functional project they can demo in interviews, and an understanding of why these models work — not just which API to call.
Curriculum
Day 1 — LLMs, APIs, and Prompt Engineering (8 hours)
Module 1: How LLMs Work (1.5 hours)
- Transformer architecture intuition (no math heavy-lifting)
- Tokenization, embeddings, and context windows
- RLHF and instruction fine-tuning — why GPT-4 follows instructions
- Comparing frontier models: GPT-4o, Claude 3, Gemini, Llama 3
Module 2: Working with the OpenAI API (2 hours)
- API authentication, rate limits, cost management
- Chat completions, system prompts, and temperature
- Streaming responses in a web app
- Hands-on: Build a domain-specific Q&A bot
Module 3: Prompt Engineering Techniques (2.5 hours)
- Zero-shot, few-shot, and chain-of-thought prompting
- ReAct prompting for reasoning + action
- Structured output with JSON mode
- Prompt injection risks and mitigation
- Hands-on: Prompt optimization lab — iterating to get reliable outputs
Module 4: LangChain & Agentic Pipelines (2 hours)
- LangChain chains, memory, and tool use
- Building a RAG (Retrieval-Augmented Generation) pipeline
- Vector databases: ChromaDB basics
- Hands-on: Build a chatbot that answers questions from a PDF
Day 2 — Multimodal AI & Deployment (8 hours)
Module 5: Image Generation with Stable Diffusion (3 hours)
- Diffusion model intuition — forward and reverse process
- Prompt anatomy for image generation
- ControlNet and image-to-image workflows
- Fine-tuning on custom subjects with LoRA (Dreambooth)
- Hands-on: Generate a consistent character / product visualization
Module 6: Hugging Face Ecosystem (2 hours)
- Model Hub — finding and loading models
- Inference API vs. local inference
- Gradio for rapid demo UIs
- Hands-on: Deploy a sentiment analysis model with a Gradio interface
Module 7: Capstone Project (3 hours)
- Teams of 3–4 build one of:
- A RAG-powered college FAQ bot
- An AI image generator for a specific domain
- A prompt-powered content repurposing tool
- Code review and demo presentation
Who Should Attend
- Final year BE / MCA students in CS, IS, AI/ML, or Data Science streams
- Students preparing for product company placements
- Anyone who has seen GenAI demos and wants to understand how to build them
AICTE Activity Points
This workshop is structured to qualify for AICTE activity points under the Professional Development category.
Inquiry
Interested in running this program for your college? Request a proposal →