Generative
AI
12 week Course Curriculum
Here’s a 3month “Generative AI” curriculum designed with weekly topics, monthly tests, assignments, and projects, aligned to 2026 industry standards and focused on beginner-to-intermediate Generative AI, covering LLMs, prompt engineering, RAG, netuning, image/code generation, AI agents, and deployment.
Assumptions :
4–5 hours per day, 5 day weeks → 12 weeks ≈ 3 months.
Tools :
Python, Jupyter/Colab, Hugging Face, OpenAI/Anthropic/Gemini APIs, LangChain/LlamaIndex, vector DBs (Chroma/FAISS), Gradio/Streamlit, Git/GitHub, optional LoRA/PEFT for netuning.
Outcome :
Prepares students for Generative AI Engineer / LLM Application Developer / AI Product Developer roles in startups, edutech, SaaS, and creative industries.
Overall Assessment Plan Weekly assignments :
Weekly assignments: 2–3 small tasks (prompts, notebooks, miniapps, API integrations).
Monthly tests: 1hour MCQ + short practical (e.g., “build a RAG pipeline” or “ netune a small model”).
Monthly projects: Endtoend GenAI applications (chatbot, RAG system, image generator, AI agent).
Month 1 – Foundations: AI/ML, LLMs & Prompt Engineering (Weeks 1–4)
Weeks & Topics | Daily Task (4-5 hrs) | Assignments | Milestones |
|---|---|---|---|
week 1 : Intro to AI, ML, DL & GenAI | AI vs ML vs DL vs Generative AI, supervised/unsupervised/reinforcement learning, neural nets basics, what GenAI can/cannot do, realworld use cases. | 1) Write a 1page comparison of AI/ML/DL/GenAI; 2) List 10 GenAI use cases across domains (health, f inance, edutech, etc.). | Short quiz on foundational concepts |
week 2 : Python for GenAI & data basics | Python refresher, working with APIs (requests), JSON, basic data handling, environment setup (Colab/VS Code, APIs keys). | 1) Call a public LLM API (e.g., OpenAI/Gemini) from Python and parse response; 2) Store outputs in a CSV/JSON file. | First LLMAPI script running. |
week 3 : LLM fundamentals & tokens | How LLMs work (transformers concept), tokens, embeddings, context window, temperature, topp, prompts vs completions, model families (GPT, Llama, Mistral). | 1) Experiment with temperature/topp and observe output differences; 2) Compare outputs from 2 different models on the same prompt. | Prompt experiment log + comparison table. |
week 4 : Prompt engineering I & II | Basic prompting, fewshot prompting, chainofthought, role prompting, unsafe prompts & ethics, Monthly Test 1 (LLM basics + prompt engineering). | 1) Design 10 prompts for different tasks (summarization, Q&A, code gen, data analysis); 2) Build a “prompt library” notebook. | Project 1: “Prompt Engineering & GenAI Toolkit”– A notebook + small CLI/Gradio app that:- Offers 5+ prompt templates (summarizer, question answerer, code generator, data explainer, content rewriter)- Uses one or more LLM APIs- Shows before/after outputs and bestpractice tips. |
Month 2 – Advanced GenAI: RAG, FineTuning & Multimodal (Weeks 5–8)
Weeks & Topics | Daily Task (4-5 hrs) | Assignment | Milestones |
|---|---|---|---|
week 5 : Embeddings & vector databases | What embeddings are, similarity search, vector DBs (Chroma, FAISS, Pinecone), indexing, retrieval basics. | 1) Generate embeddings for a small text corpus; 2) Store in a vector DB and run similarity search. | Vector DB demo + similarity results. |
week 6: Retrieval Augmented Generation (RAG) | RAG architecture, document loading, chunking, retrieval + generation pipeline, evaluation of RAG, basic RAG with LangChain/LlamaIndex | 1) Build a RAG pipeline over your own PDFs/Docs; 2) Compare RAG vs plain LLM on domain questions. | RAG notebook + evaluation notes. |
week 7 : Finetuning & PEFT (LoRA) | Full finetuning vs LoRA/QLoRA/PEFT, dataset preparation, training loops (concept), Hugging Face Trainer, deployment of finetuned models (concept). | 1) Prepare a small instruction tuning dataset; 2) Finetune a small model (e.g., Llama27Bchat or smaller) using LoRA on Colab | Finetuning notebook + before/after comparison. |
week 8 : Image & code generation, multimodal | Diffusion models (Stable Diffusion), image generation & editing, code generation (Codexstyle), multimodal LLMs (text+image), safety & bias, Monthly Test 2 (RAG + finetuning + multimodal). | 1) Generate images with prompts and edit them; 2) Use an LLM to generate & explain code snippets for a small task. | Project 2: “Domain RAG Assistant” – A RAGbased chatbot for a specific domain (e.g., your robotics/edutech startup docs, course material, or a public dataset) with:- Document ingestion (PDFs/Markdown)- Chunking + vector DB- LLM + RAG pipeline- Simple UI (Gradio/Streamlit)- Evaluation on 10–15 test questions. |
Month 3 – AI Agents, Automation, Deployment & Capstone (Weeks 9–12)
Weeks & topics | Daily tasks (4-5 hrs) | Assignment | Milestones |
|---|---|---|---|
week 09 : AI agents & tool use | Agents concept, role of LLM as “brain”, function calling, tool use (search, API calls, code execution), multistep planning, simple agent frameworks. | 1) Build a simple agent that can: search (mock or real), call an API, and produce a summarized answer; 2) Add memory (session history). | Agent demo notebook + flow diagram. |
week 10 : AI automation & workflows | Nocode automation (Zapier/Make), eventbased workflows, connecting LLMs to business tools (Slack, email, CRM), AI safety & guardrails. | 1) Design an automation workflow (e.g., “new lead email → summarize → add to sheet → notify Slack”); 2) Implement a simplified version with APIs. | Automation design doc + prototype. |
week 11 : Deployment, MLOps & monitoring | Deploying GenAI apps (Gradio/Streamlit to Hugging Face Spaces, Docker, cloud), logging, usage tracking, cost control, basic monitoring & feedback loops. | 1) Deploy your RAG assistant or agent to a public URL; 2) Add basic logging & usage metrics. | Live deployed app + monitoring snippet. |
week 12 : Capstone & portfolio | Endtoend GenAI product: problem → data → model/RAG/agent → UI → deploy → present; Final Test (MCQ + practical capstone review). | 1) Polish capstone; 2) Write README, 2–3 page report, and prepare a short demo video + slides. | Project 3: “Generative AI Capstone” – A complete, portfolioready GenAI system such as:- “Companyspecific RAG chatbot” for your startup’s docs/courses- “AI tutor agent” that answers student questions using your curriculum + web data- “Automated content generator” (blog/quiz/lesson planner) with human in the loop review Includes: data ingestion, RAG or finetuned model, agent/workflow logic, UI (Gradio/Streamlit), deployment, and a brief security/ethics note |
