There are no separate tracks to choose between, and no risk of picking wrong. There is one carefully sequenced path that every graduate walks together, beginning with the fundamentals and ending with a real, deployed AI product in your hands. Here is exactly what that journey looks like.
Every module rests on the one before it, so you never feel lost and nothing later arrives as a surprise. You start by truly understanding how language models work, then learn to guide them, give them knowledge, let them act as agents, and finally ship a real application you can be proud of.
No prior coding experience is required, and the program is offline, so you learn hands-on, alongside mentors and fellow graduates, from the very first day to the final demo.
We start at the foundation, so nothing later feels like magic. You learn how transformers and attention work, what tokens and embeddings really are, how models are pretrained and fine-tuned, what context windows and sampling do, and why models sometimes hallucinate. By the end you can reason about model behaviour with genuine understanding.
You learn to speak to models with precision: prompt patterns, chain-of-thought, few-shot prompting, structured and JSON outputs, and tool and function calling. You build directly against the OpenAI and Anthropic Claude APIs, and learn to manage cost, latency and reliability the way professionals do.
You give AI access to real, trustworthy knowledge. You build RAG pipelines, explore GraphRAG and hybrid search, and work with vector stores including Pinecone, Weaviate, ChromaDB, FAISS and pgvector, plus long-term memory tools such as Graphiti, mem0 and Zep.
This is where AI begins to act on its own. You build agents and multi-agent systems using LangChain, LangGraph, CrewAI and AutoGen, mastering tool use, memory and planning, and working with the Anthropic Claude API, the OpenAI Agents SDK, Google ADK (Agent Development Kit) and the Model Context Protocol (MCP).
You turn everything into a real product. You design full-stack GenAI applications with streaming and guardrails, add evaluation and observability, and deploy using FastAPI, Streamlit and Docker, always mindful of cost, latency and reliability in production. You finish with a working AI application of your own, ready to show at any interview, including at Bonami Software.
Career paths this prepares you for:
One path, walked together, from the fundamentals to a real AI product in your hands. If you are ready to start, we would be glad to walk it with you.