Welcome to the comprehensive 10-week course on building production-ready AI agents and mastering the entire ML engineering lifecycle in the era of Large Language Models (LLMs).
This course integrates every key topic in modern AI engineering: Data Engineering, foundational LLM concepts, Retrieval-Augmented Generation (RAG), LLM fine-tuning, and model alignment. Everything builds toward creating your own end-to-end research agent that can search papers, extract content via OCR, generate summaries, and even create podcast-style content.
- Duration : 10 weeks
- Format : 2-hour weekly lectures + hands-on projects
- Focus : Data Engineering for LLMs
- Final Deliverable : Deployable AI research agent
| Week | Topic | Key Concepts | Project |
|---|---|---|---|
| Week 1 | Intro to LLMs & Prompt Engineering | Generative AI, prompting techniques (CO-STAR), JSON/XML output | Design prompts for research agent using CO-STAR framework |
| Week 2 | LLM Architecture & Training | Transformers, hallucination, SFT/DPO/PPO, scaling laws | Run local LLM inference (LLaMA 3/4), evaluate with custom prompts |
| Week 3 | Data Collection & Extraction | Web scraping, OCR (Tesseract/Surya), ASR (Whisper), data cleaning | Scrape arXiv, OCR PDFs, filter & clean data for pretraining |
| Week | Topic | Key Concepts | Project |
|---|---|---|---|
| Week 4 | Retrieval-Augmented Generation (RAG) | Embeddings, chunking, vector DBs, LangChain | Build RAG pipeline to augment LLM with external knowledge |
| Week 5 | Supervised Fine-Tuning (SFT) I | Full vs. LoRA fine-tuning, ChatML format, TRL/Deepspeed | Apply LoRA and full fine-tuning, explore overfitting |
| Week 6 | Supervised Fine-Tuning (SFT) II | Synthetic data, quality checks, LLM-as-judge | Generate synthetic SFT data, perform ablation studies |
| Week 7 | Model Alignment | RLHF, DPO/PPO, reward modeling, data labeling | Build Gradio labeling tool, run DPO alignment experiment |
| Week | Topic | Key Concepts | Project |
|---|---|---|---|
| Week 8 | Safety & Ethics | Hallucination prevention, jailbreak methods, bias mitigation | Test model safety, explore jailbreaking, safety datasets |
| Week 9 | Voice & Multimodal AI | GPT-4o real-time, ASR/TTS pipelines, chained agents | Build voice agent (GPT-4o style), explore NotebookLM pipeline |
| Week 10 | Final Capstone | Agents, MCP protocol, function calling, task chaining | Complete end-to-end research agent with voice output |
Build an intelligent agent that enables natural language queries about research papers:
- Data Engineering : Collect and preprocess academic papers
- RAG Integration : Ground responses in real research documents
- Fine-Tuning : Personalize with supervised fine-tuning
- Alignment : Ensure safe, accurate, and relevant answers
- Deployment : Create a working, demonstrable agent
Develop a voice-driven research assistant:
- Integrate ASR, LLMs, and TTS
- Build chained audio-AI pipelines
- Create demo video for portfolio
Design an agent aligned with your career interests:
- Choose any domain (music, biotech, legal, etc.)
- Present at public showcase event
- Compete for top project recognition
- LLMs : LLaMA 3/4, ChatGPT, Claude
- Frameworks : LangChain, TRL, Deepspeed
- Data : Web scraping, OCR (Tesseract/Surya), ASR (Whisper)
- Vector DBs : For RAG implementation
- Fine-tuning : LoRA, full fine-tuning methods
- MCP (Model Context Protocol) : For agent integration
- Gradio : For building labeling interfaces
- Git : Version control and collaboration
- Discord : Community Q&A and code sharing
By the end of this course, you will:
✅ Master Modern AI Engineering : From data collection to model deployment
✅ Build Production-Ready Agents : Complete end-to-end AI systems
✅ Understand LLM Lifecycle : Pretraining, fine-tuning, and alignment
✅ Implement RAG Systems : Advanced retrieval-augmented generation
✅ Deploy Real Applications : Career-ready portfolio projects
✅ Navigate AI Safety : Ethical considerations and safety alignment
- Python 3.8+
- Basic understanding of machine learning
- Familiarity with Git and command line
- 8GB+ RAM (16GB recommended for local LLM inference)
# Clone the repository
git clone https://github.com/inference-ai-course/MLE_in_Gen_AI-Course.git
cd MLE_in_Gen_AI-Course
# Create virtual environment
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
# Install dependencies
pip install -r requirements.txt
# or just follow the instruction from the jupyter Notebooks
- API Keys : Set up OpenAI, Anthropic, or other LLM API keys
- MCP Setup : Configure Model Context Protocol for agent integration
- Discord : Join the course Discord for Q&A and collaboration
| Week | Milestone | Deliverable |
|---|---|---|
| 1 | Project Kickoff | Define research agent goals, initial prompts |
| 4 | Project Insight I | Share progress, receive peer feedback |
| 7 | Project Insight II | Lock in project direction & components |
| 10 | Final Presentation | Working agent demo, learnings, technical depth |
- Discord Server : Real-time Q&A and code sharing
- Office Hours : Weekly TA sessions for project guidance
- Peer Review : Collaborative feedback sessions
- Showcase Event : Public presentation of final projects
- Deployable Research Agent : Live demo for interviews
- GitHub Portfolio : Complete, documented projects
- Technical Blog Posts : Document your learning journey
- Demo Videos : Showcase multimodal agent capabilities
- Top Project Awards : Judged showcase competition
- Industry Connections : Guest speakers and networking
- Open Source Contributions : Contribute to course materials
- Minimum : 8GB RAM, modern CPU
- Recommended : 16GB+ RAM, GPU for local training
- Cloud Alternative : Google Colab Pro, AWS, or similar
- Python 3.8+, Node.js (for MCP)
- Git, Docker (optional)
- Code editor (VS Code recommended)
- Week 1 : Head to
week01/and follow the setup instructions - Join Discord : Connect with classmates and instructors
- Define Your Goal : Write your one-sentence agent mission
- Start Building : Begin with prompt engineering fundamentals
Your AI engineering journey starts now! 🚀
This course is designed to be highly practical, career-focused, and immediately applicable to real-world AI engineering roles. Every project builds toward creating tangible, demonstrable skills that will set you apart in the rapidly evolving AI landscape.