What you'll learn
- Call frontier models (OpenAI, Claude, Gemini) and run open-source models locally with Ollama
- Build chat UIs with streaming, tools and multi-modal input
- Use Hugging Face pipelines, tokenizers and quantised models
- Pick the right model using benchmarks, cost and latency
- Build RAG systems with embeddings, vector databases and LangChain
- Fine-tune frontier models and open-source models with LoRA / QLoRA
- Evaluate models properly against a baseline
- Ship an autonomous multi-agent capstone
Curriculum
8 modules · 28 topics · hands-on labs in every module
Week 1: Your First LLM Application4 topics
- Setting up your AI engineering environment
- Frontier model APIs vs local models with Ollama
- Prompting: system vs user prompts
- Project: AI website summariser
Week 2: Frontier APIs & User Interfaces4 topics
- Streaming responses
- Building UIs with Gradio (and Next.js for Node track)
- Tool calling and multi-modal chat
- Project: customer-support assistant with tools
Week 3: Open-Source with Hugging Face3 topics
- Pipelines, tokenizers and models
- Quantisation and running on GPUs (Colab)
- Project: meeting-minutes generator from audio
Week 4: Choosing the Right LLM3 topics
- Benchmarks and leaderboards
- Cost, latency and context-window trade-offs
- Project: code-conversion tool using LLMs
Week 5: Retrieval-Augmented Generation (RAG)4 topics
- Embeddings and vector databases (Chroma)
- Chunking strategies and retrieval
- LangChain / LangChain.js
- Project: company knowledge-base assistant
Week 6: Fine-Tuning Frontier Models4 topics
- Data curation and cleaning
- Baselines with traditional ML
- Fine-tuning through APIs and evaluation
- Project: product price predictor
Week 7: Fine-Tuning Open-Source Models3 topics
- PEFT, LoRA and QLoRA explained
- Training runs and experiment tracking
- Project: beat a frontier model with your own fine-tune
Week 8: Capstone – Autonomous Agent System3 topics
- Multi-agent architecture
- Serverless deployment of models
- Capstone: deal-hunting agent that notifies you
Requirements
- Basic Python or JavaScript
- No ML background needed: we teach what you need
- A laptop; GPU work runs on free cloud notebooks
About this program
This is Part 1 of the AI Engineering program and is inspired by the best-in-class LLM engineering curriculum taught worldwide. It is taught by an engineer who previously led work at LangChain.
Every week ends with something you built. By the end you will have 8 portfolio projects and a working understanding of how LLM products are really engineered.
After this program you can join the 6-month internship and take part in the ownership model: build an Incred vertical and earn a share, sell your product locally, or pitch your own idea for funding.
Your mentor
Amit Singh
CTO, Incred Applications Pvt Ltd · IIT Delhi alumnus
Leads the global engineering team of incred.golf across Germany, London, Croatia, India and Denver. Ex-lead at LangChain; previously with NVIDIA and IBM.