AI Pioneers — Cohort 1 · 2026

Build real
GenAI apps
in 8 weeks

A hands-on, practitioner-led programme for IT professionals ready to move from prompt-tinkering to shipping production RAG systems, agents, and deployed AI applications.

8wk

Live sessions

Notebooks

1×

Deployed app

∞

Carry-on skills

Join Cohort 2 → View curriculum Book a call

rag_app.py — AI Pioneers Week 7

$ python rag_app.py --query "VPN keeps dropping"

🔎 Retrieving top-4 chunks from ChromaDB...

[runbook_vpn.md] dist=0.1241

[runbook_vpn.md] dist=0.1893

🤖 Generating grounded answer with Claude...

✅ Answer:

• Check GlobalProtect logs at PanGPS.log

• Set keep-alive interval to 20 seconds

• Verify MTU is set to 1400

SOURCES: runbook_vpn.md

What you will build

From zero to production AI — in 8 weeks

Every week ships a working deliverable. By week 8 you have a live RAG application, a public GitHub portfolio, and a LinkedIn post to show for it.

🐍

Python for AI — from scratch

Virtual environments, file I/O, API calls — everything you need and nothing you don't. No math degree required.

✍️

Production prompt engineering

Zero-shot, few-shot, CoT, JSON output, IT ticket classifier with qualitative evaluation on 10 real cases.

🔗

LangChain memory chatbot

Multi-turn domain chatbot with 8 provider support — Anthropic, OpenAI, Groq, Gemini, Ollama and more.

🔍

Semantic search engine

10+ domain documents indexed in ChromaDB with embeddings. Retrieve by meaning, not by keyword.

🏗️

Full RAG pipeline

Chunk → embed → retrieve → generate with source citations. The architecture behind every enterprise AI assistant.

🚀

Live deployed AI app

Streamlit RAG chatbot deployed on Hugging Face Spaces. A real public URL to share with your team and on LinkedIn.

8-week curriculum

Week-by-week breakdown

Two 90-minute live sessions per week. Every session ends with a runnable deliverable. Click any week to see what's inside.

WEEK 01 Python Basics & AI Landscape ▾

Phase 1 — Foundations

Python crash course — variables, loops, functions, file I/O, log analyser lab

AI landscape + first Anthropic API call — tokens, context, temperature

Working Python script that calls Claude API and handles missing key gracefully

WEEK 02 Prompt Engineering ▾

Phase 1 — Foundations

Prompt anatomy — instruction, context, format, constraints, few-shot

CoT, structured JSON output, IT ticket classifier lab with evaluation

Notebook with 3 real-work prompts + IT ticket classifier evaluated on ≥10 cases

WEEK 03 AI APIs & LangChain Intro ▾

Phase 2 — APIs & Tools

Auth patterns, streaming, multi-turn history, token management

LangChain LCEL chains, RunnableWithMessageHistory, memory chatbot

Domain-specific memory chatbot — 8-provider support, 6+ conversation turns

WEEK 04 Embeddings & Vector Databases ▾

Phase 2 — APIs & Tools

Embeddings intuition — GPS analogy, cosine similarity from scratch

ChromaDB — collections, upsert, query, metadata filters

≥10 domain documents in ChromaDB + semantic search with 5 queries documented

WEEK 05 Building a RAG Pipeline ★ ▾

Phase 3 — RAG Systems

RAG architecture — chunk → embed → retrieve → augment → generate

Ingest pipeline, retrieval function, grounded generation with citations

Full RAG chatbot over ≥5 real documents with source citations on every answer

WEEK 06 Advanced RAG & Evaluation ▾

Phase 3 — RAG Systems

Re-ranking, hybrid search (BM25 + dense), query rewriting, HyDE

LLM-as-judge evaluation, golden Q&A set, before/after comparison table

Evaluated RAG system with documented quality metrics and top-2 failure modes

WEEK 07 AI Agents & Deployment ▾

Phase 4 — Deploy & Ship

Agent loop — tool calling, plan/act/observe, max steps guard

Streamlit UI, Hugging Face Spaces deployment, secrets management

Live, publicly accessible AI web application with working RAG backend

WEEK 08 Capstone & Demo Day ▾

Phase 4 — Deploy & Ship

Capstone build — project spec, peer review, polish sprint

Live 5-minute demos, career tracks, Layer 3 roadmap

Polished capstone repo + LinkedIn post + demo video

Technology stack

Tools you will use in production

No toy frameworks. Every tool is either already used in enterprise AI systems or on the path there. Local-first with Ollama — no cloud API key required to get started.

🐍

Python 3.11

Core language

🔗

LangChain

Orchestration

🗄️

ChromaDB

Vector store

🤖

Anthropic Claude

LLM API

🦙

Ollama

Local LLMs

📊

Streamlit

Web UI

🤗

Hugging Face

Models + Deploy

🔭

LangSmith

Observability

📄

pypdf

Document ingest

🔮

sentence-transformers

Embeddings

☁️

OpenAI / Groq / Gemini

Multi-provider

💻

VS Code / Cursor

IDE

RAG Architecture Semantic Search Vector Embeddings LangChain LCEL Memory & History Streaming APIs Chunking Strategies Re-ranking LLM-as-Judge Eval HF Spaces Deploy Agent Tool Calling Source Citations

Cohort 1 — in progress

What students are saying

Cohort 1 is live through Week 5. Testimonials will be added as students complete their capstone projects in Week 8.

"Coming from a network engineering background with zero Python experience, I was sceptical I could build a RAG app in 8 weeks. By Week 5 I had a working chatbot answering questions from my own runbooks with citations."

Arun K.

Network Engineer · Cohort 1

"Mayank's IT analogies made embeddings and vector databases click immediately. 'ChromaDB is PostgreSQL for embeddings' — I finally understood what I was building and why, not just how to run the code."

Sonal R.

IT Operations Lead · Cohort 1

📝

More testimonials coming after Demo Day — Week 8

Your instructor

Built by a practitioner, taught by a practitioner

Mayank Chugh

Azure OpenAI & GenAI Architecture
20+ years enterprise experience

📺 YouTube @itaienthusiast 💻 GitHub 💼 LinkedIn ✍️ Medium 📅 Book a 1:1 call 🛒 Course page

Enterprise AI — delivered in plain English

Mayank Chugh is an AI Solutions Architect with 20+ years of enterprise experience, including production GenAI deployments at Cathay Pacific and HSBC. He has shipped RAG pipelines, agentic systems, and LangChain integrations in financial services environments where hallucination is not an option and provenance is a compliance requirement.

He runs the IT AI Enthusiast YouTube channel — a technical education channel targeting AI/ML practitioners and enterprise architects — and produces content on Advanced RAG, LangGraph, Azure OpenAI, and production GenAI patterns.

AI Pioneers is the course Mayank wished existed when he started — application-first, top-down, no unnecessary theory. Every session builds something runnable. Every IT analogy comes from real enterprise experience, not a textbook.

Production RAG delivery at Cathay Pacific and HSBC
Advanced RAG Series — Hybrid Search, Reranking, Evaluation (YouTube)
12+ certifications: AZ-400, AWS Solutions Architect, Terraform, TOGAF, CrewAI
MCP server ecosystem — 5 public GitHub repos
HireIQ AI recruitment platform (LangGraph state machine)

AZ-400 AWS SA Terraform TOGAF CrewAI Azure OpenAI LangChain LangGraph

Build real GenAI apps in 8 weeks