Batch 1 — Enrolling Now

Master Generative AI by Building on a Real AI Platform

A 10-week hands-on cohort covering LLMs, RAG, AI Agents, and Fine-Tuning. Not just videos — you build on our live Jupyter-powered LLM platform.

Next cohort starts March 2026 · Limited to 25 seats
750+
Professionals Trained
10
Weeks, Hands-On
18+
Portfolio Projects
100%
Practical, No Fluff
AGAI-101

Applied GenAI: LLMs, RAG & Agents

Go from zero GenAI knowledge to deploying production-grade AI applications. 10 weeks, 8 projects, 1 capstone — everything you need to build with LLMs.

View full curriculum →
🧠

Deep Understanding

Transformer internals, self-attention, tokenization, and the full GPT training pipeline — not surface-level tutorials.

🔧

Build Real Systems

RAG pipelines, multi-agent workflows, fine-tuned models. Every week includes a real-world workshop using live data from APIs like Hacker News, USGS Earthquakes, and GitHub.

🚀

Production Ready

Security guardrails, FastAPI deployment, Docker containerization. Ship GenAI apps that are actually safe to deploy.

Our Platform

Learn on a Live AI Platform, Not Just Videos

Most courses give you pre-recorded videos and Colab notebooks. We built a Jupyter-backed environment with note cells, AI prompt cells, and code cells — where AI acts as your always-on instructor. It sees your work, understands your context, and helps you learn in real time.

  • AI-aware cells — the AI sees all cells above, your code, notes, and prompts for full context
  • Built-in sandbox — AI can execute and test code in an isolated environment before affecting your kernel
  • AI with tools — system prompt, code execution, context from your entire notebook at its fingertips
  • Always-on AI instructor — solving a workshop or assignment? Get instant help, like having the instructor beside you 24/7
  • 2 workspaces & 10 notebooks each — enough room for coursework, experiments, and side projects
  • Need more power? Top up credits anytime for access to stronger models or additional tokens for your projects
bsigma-lab.ipynb
# AGAI-101 | Week 5: RAG Pipeline

from langchain import ChatOllama
from chromadb import Client

# Load your documents
docs = load_pdfs("./papers/")
chunks = split(docs, chunk_size=512)

# Build the RAG chain
retriever = vectorstore.as_retriever()
chain = retriever | llm | parse

chain.invoke("Summarize the key findings")
# → "The paper presents three main findings..."
10-Week Curriculum

From Foundations to Full-Stack GenAI

Each week combines theory, hands-on coding, and a portfolio project. You graduate with 8+ GitHub projects.

Phase 1LLM Foundations
1LLM Foundations & Your First AI Interaction

Topics: How LLMs work, attention mechanism, model capability ladder, reasoning models, provider landscape, API anatomy, tokens, context windows, temperature, top-p

Project: Model Comparison Tool — same prompt, 3 models, side-by-side with quality assessment

Real-World Workshop: AI News Intelligence Dashboard — live Hacker News data, AI categorization, audience-adapted briefings, token economics

2Prompt Engineering & Solving Real NLP Tasks

Topics: Zero-shot, few-shot, chain-of-thought, prompt mechanics, constrained decoding, multi-provider structured output, prompt evaluation

Project: Customer Review Analyzer — raw reviews to structured JSON with multi-provider comparison and evaluation metrics

Real-World Workshop: AI Recipe Transformer — live TheMealDB data, dietary adaptation, cuisine fusion, CoT meal planning

Phase 2RAG — Retrieval-Augmented Generation
3Embeddings, Vector Search & RAG Fundamentals

Topics: Contrastive learning, MTEB benchmark, embeddings, semantic search, ChromaDB, ANN/HNSW, RAG pipeline, chunking, principled K selection, distance thresholding

Project: Company Knowledge Base Bot — RAG with confidence-aware retrieval and distance thresholding

Real-World Workshop: AI Book Recommendation Engine — Open Library data, semantic search over real books, metadata filtering

4Building a Production RAG Application

Topics: Pipeline pattern, pure-Python RAG, document loaders, chunking strategies, retrieval failure taxonomy, hybrid search (BM25 + semantic), distance thresholding, conversational RAG

Project: Document Q&A System — multi-document RAG with citations, hybrid search, threshold calibration, failure diagnosis

Real-World Workshop: AI Country Intelligence Briefing — REST Countries + Open-Meteo, multi-source RAG, conversational retrieval

5RAG Evaluation & Optimization

Topics: RAGAS metrics, LLM-as-Judge, judge validation (Cohen's kappa), statistical rigor (confidence intervals), cost/latency tracking, query expansion, HyDE, debugging RAG failures

Project: RAG Evaluation Pipeline — baseline scores, optimization experiments, judge validation, statistical rigor, cost/latency tracking

Real-World Workshop: AI Earthquake Analysis — live USGS seismic data, eval dataset creation, retrieval metrics, statistical rigor

Phase 3AI Agents
6AI Agents & Function Calling

Topics: Agents vs chatbots, function calling (OpenAI + Anthropic), multi-provider abstraction, agent loop, tool design principles, agent safety (cost budgets, loop detection), error handling

Project: Personal Assistant Agent — 6 tools, cost tracking, safety guards, multi-provider comparison

Real-World Workshop: AI Space Tracker Agent — live ISS + Sunrise-Sunset + weather APIs, real-time tool calling, multi-tool agent

7LangGraph & Agentic Workflows

Topics: Graph abstraction, pure Python executor, LangGraph, state management, conditional routing, self-reflection (Reflexion), graph anti-patterns, human-in-the-loop, agent memory

Project: Content Pipeline Agent — research, draft, review cycles with self-reflection and graph design patterns

Real-World Workshop: AI Content Publishing Pipeline — Wikipedia + HN + Quotable, LangGraph editorial workflow, self-reflection

8Multi-Agent Systems & MCP

Topics: Multi-agent architectures, coordination strategies (supervisor/debate/consensus), MCP protocol mechanics (JSON-RPC), multi-agent failure modes, agent safety guards

Project: Multi-Agent NovaTech System — supervisor + specialized workers + MCP tool server with failure recovery

Real-World Workshop: AI OSINT Intelligence Team — GitHub + HN + Wikipedia APIs, multi-agent intelligence, MCP server

Phase 4Fine-Tuning, Security & Capstone
9Fine-Tuning & LLM Security

Topics: Fine-tuning pitfalls (catastrophic forgetting, cost-benefit), LoRA mechanics (low-rank decomposition, rank selection), QLoRA, Unsloth, alignment as security (RLHF, Constitutional AI), guardrails frameworks, prompt injection defense, OWASP Top 10 for LLMs

Project: Fine-tune a model + add guardrails with forgetting checks, held-out validation, and red team exercise. Capstone kickoff.

Real-World Workshop: AI Security Red Team Lab — real CVE data from NIST, layered defenses, automated red team attacks

10Capstone Project & Graduation

Topics: System composition, framework independence (architecture-first thinking), advanced orchestration (self-reflection, parallel, HITL), production deployment (FastAPI, monitoring, observability), Demo Day presentations

Project: NovaTech AI Assistant capstone — self-reflection workflows, framework-independent architecture, production deployment sketch

Real-World Workshop: Build Your Own AI Startup MVP — choose a real problem, combine all course techniques, investor pitch

Your Instructor

Bala

Sr. Software Engineer & AI Practitioner

10+ years building production AI systems for global enterprises including major FMCG brands like Coca-Cola, Pepsi, GSK, and Sanofi. Former Associate Lecturer at the National University of Singapore, where he trained 750+ professionals and won the Teaching Excellence Award.

🏆NUS Teaching Excellence Award
AWS Solutions Architect
🧠Google TensorFlow Developer
📚Deep Learning Specialization
Led AI backend in Retail AI Space — serving almost all major FMCG brands Pepsi, Coca-Cola
Trained 750+ professionals at NUS Singapore
Built ML inference pipelines & image recognition engines
Expert in Python, FastAPI, Docker, K8s, AWS, GCP
Built a Jupyter-powered LLM platform from scratch

Who Is This For?

Software Engineers

You write code daily but want to add GenAI to your skillset. You want to build with LLMs, not just use ChatGPT.

Data Scientists & ML Engineers

You know classical ML but need to bridge the gap to LLMs, RAG, agents, and fine-tuning.

Tech Leads & Architects

You need to evaluate and architect GenAI solutions for your organization. Understand what's hype vs what works. Each week includes production-tier challenges.

Career Switchers

You have programming fundamentals and want to break into the hottest area of tech with a real portfolio.

Pricing

Invest in Your GenAI Career

MOST POPULAR

Live Cohort

AED 2,999 / person

~USD 817 · One-time payment

  • 10 weeks of live sessions (weekends)
  • Full access to bsigma LLM platform
  • 8+ guided portfolio projects
  • Direct access to instructor
  • Certificate of completion
  • Lifetime access to recordings
  • Private community access
Join Waitlist
FOR TEAMS

Corporate Workshop

Custom

2-3 day intensive · On-site or virtual

  • Customized to your team's tech stack
  • Hands-on labs on bsigma platform
  • Use cases tailored to your industry
  • Up to 20 participants per workshop
  • Post-workshop support & resources
  • Certificate for all participants
  • Invoice & PO supported
Contact Us

Frequently Asked Questions

What are the prerequisites?
You should be comfortable with Python programming and have basic understanding of machine learning concepts. Prior experience with neural networks is helpful but not required.
What if I miss a live session?
All sessions are recorded and available within 24 hours. You also get lifetime access to all recordings after the cohort ends.
Do I need a GPU or expensive hardware?
No. All hands-on work runs on our bsigma platform, Google Colab (free tier), or Kaggle notebooks. No local GPU required.
What LLMs will we use?
We primarily use open-source models (Llama, Mistral, Gemma) via Ollama and HuggingFace, plus free tiers of commercial APIs for comparison.
Will I get a certificate?
Yes. Upon completing the cohort and your capstone project, you receive a certificate of completion from bsigma.ai.
Is there a refund policy?
If you are not satisfied after the first 2 weeks, we offer a full refund, no questions asked.

Join the Waitlist

Batch 1 starts March 2026. Limited to 25 seats. Get notified when enrollment opens.

No spam. We only email when enrollment opens.