Master the tools, frameworks, and cloud AI stack that modern engineers work with — Python, LLM APIs, LangChain, LangGraph, LlamaIndex, CrewAI, AutoGen, MCP, RAG, LangSmith, and AWS / Azure / GCP AI foundations. Every week ends with a hands-on project deliverable.
Next cohort starts 16 August 2026 · Live sessions: 8:00 AM – 11:00 AM IST · Founding-cohort pricing · Regular price ₹24,999 + GST.
A short note from Nachiketh on why most Agentic AI learners get lost — and how this bootcamp fixes it.
Every week there's a new framework, a new agent pattern, a new cloud AI service. Ten scattered tutorials on YouTube. Five different opinions on which stack to learn first. Twenty tools that look important but aren't obviously connected.
Prompt-only tutorials that stop at "chat with the model" and never reach real agent architecture.
Flagship production programs entered before the underlying tools and frameworks are second nature.
Disconnected weekend courses that don't add up to a working developer skill set.
The Agentic AI Developer Bootcamp assembles the essential tools, frameworks, and cloud services into a coherent developer skill set — and gives you the project artifacts to prove it.
You leave with real fluency — the kind you can talk about in an engineering conversation, and the kind you can show in code.
By the end of the bootcamp, you have both the skill and the artifacts to demonstrate it.
Build agents that call tools, structure outputs, and handle real errors — not chat wrappers.
Wire tool schemas and function calls with judgment — expose the right surface, hide the wrong one, and catch failure gracefully.
Use Claude Code, Cursor, and GitHub Copilot as core developer tools — the way senior AI engineers work every day.
Build RAG assistants with LangChain and LlamaIndex — chunking, embeddings, retrievers, and the failure modes you have to see coming.
Design graph-based agent workflows in LangGraph — state, conditional routing, memory, and human-in-the-loop.
Compare and use multi-agent frameworks — CrewAI, AutoGen, Agno / Phidata — and know exactly when each fits.
Work with the Model Context Protocol (MCP) at real developer depth — client / server, tools, resources, prompts.
Navigate cloud AI foundations across AWS Bedrock, Azure OpenAI, and GCP Vertex AI.
Add tracing, evaluation, and observability to workflows with LangSmith and Langfuse.
Ship six weekly hands-on artifacts + one final portfolio project — every one of them in your GitHub.
Speak the language of Agentic AI development fluently — frameworks, patterns, evaluation, cloud, deployment.
If you're already writing code and curious about AI, you'll thrive here.
Ready to add Agentic AI to your core skill set.
Moving into AI-native workflows and tooling.
Looking for a serious, structured way into the modern AI stack.
Wanting a real technical foundation in Agentic AI.
Read about LangChain, CrewAI, and MCP but haven't built anything real yet.
Want to hold their own in a modern AI engineering conversation — with the projects to back it up.
Two pre-bootcamp Python classes get everyone ready. Six weekly modules take you through the full stack.
Setting the developer environment right before we build.
Python setup, VS Code and Cursor setup, virtual environments, variables, functions, lists, dictionaries, JSON handling, file handling, error handling, basic project structure.
Build a Python script that reads a file, extracts key information, and saves structured JSON.
The developer discipline behind every serious AI application.
HTTP basics, API calls, API keys and secrets, environment variables, .env files, JSON request/response payloads, Python SDK usage, Git and GitHub basics, Claude Code and Cursor workflow awareness.
Build a CLI application that calls an LLM API and returns structured output.
The mental model, the vocabulary, and your first shipped assistant.
Agentic AI vs Generative AI, agents vs workflows, the autonomy spectrum, core agent components, LLM API fundamentals, prompting design, system prompts, user prompts, structured outputs, JSON response engineering, Cursor / Claude Code / GitHub Copilot as first-class developer tools.
Build a working AI assistant using an LLM API with structured output.
The layer that turns a chat model into a real agent.
Tool calling and function calling in depth, tool schema design, tool execution flow, tool error handling, common misuse risks, the Model Context Protocol (MCP), MCP client / server concept, tools, resources, and prompts inside MCP, where MCP fits in modern Agentic AI systems.
Build a tool-using assistant with a calculator, file-summary tool, mock search tool, and full error handling.
The retrieval spine underneath every serious AI application.
Document loading, chunking strategy, embeddings, vector stores, retrievers, the full RAG pipeline, LangChain and LCEL, LlamaIndex, framework comparison, RAG failure modes and how to spot them early.
Build a document Q&A assistant using LangChain or LlamaIndex.
The graph-based agent workflow model behind modern production systems.
Why LangGraph exists, graph-based workflows, nodes and edges, state schemas, conditional routing, short-term memory, long-term memory patterns, human-in-the-loop, agent workflow design principles.
Build a LangGraph workflow with state, routing, tool call, memory, and a conditional branch.
When you scale from one agent to many — and when you shouldn't.
CrewAI, AutoGen, Agno / Phidata, deep framework comparison, role-based agents, conversation-based agents, workflow-based agents, when multi-agent systems shine, when they become overkill, low-code / no-code AI tools — Langflow, Flowise, Relevance AI-style workflows.
Build a multi-agent research workflow with researcher, writer, and reviewer agents.
Recreate a similar workflow inside a low-code / no-code tool to see the trade-offs.
The layer that separates a demo from a system you can trust to run.
AWS AI foundations + Amazon Bedrock, Azure AI foundations + Azure OpenAI, GCP AI foundations + Vertex AI, deployment options, FastAPI awareness, Docker basics, LangSmith, Langfuse, AgentOps awareness, agentic evals fundamentals, trace-based debugging, golden dataset basics, LLM-as-judge basics, cost / latency / reliability / safety checks.
Instrument an existing agent or RAG workflow with tracing, a simple evaluation dataset, a response quality check, a tool failure check, and cost / latency observation.
Intentional coverage, not a tag cloud. Every category maps to a real Agentic AI engineering discipline.
At Manifold, no week ends without an artifact you can point to. By the end of six weeks, you have a working portfolio.
An LLM-backed assistant that returns clean, structured JSON on every call.
An agent that calls tools, handles errors, and works inside MCP concepts.
Document Q&A over your own knowledge source with LangChain or LlamaIndex.
A graph-based agent with state, memory, routing, and conditional flow.
Researcher, writer, and reviewer agents collaborating on a real task.
One of your builds instrumented with tracing, evals, and observability.
Design your own project. Submit as homework. Add it to your portfolio.
Every essential tool, framework, and cloud foundation placed inside a coherent 6-week curriculum. No hunting. No fragmentation.
You know when LangChain beats LlamaIndex, when LangGraph beats CrewAI, when MCP beats a plain tool call. Opinions grounded in practice.
No week ends without something in your GitHub. Your portfolio grows every session.
AWS Bedrock, Azure OpenAI, and GCP Vertex AI treated as first-class foundations — not afterthoughts.
LangSmith, Langfuse, tracing, golden datasets, LLM-as-judge — the discipline that separates real engineers from demo builders.
You're not watching a course. You're inside a cohort.
MCP, LangGraph, multi-agent orchestration, cloud AI, and evals — covered at real working depth. Nothing skipped. Nothing dumbed down.
The Developer Bootcamp is a new program — taught in the same live weekend-cohort format, by the same instructor, as the flagship bootcamp these engineers completed.
"I used to see an AI agent as just 'an LLM with a tool'. Now I think in terms of systems — explicit conversation memory, structured logging, and targeted error handling so failures are expected, traced, and contained. The biggest unlock was separating capability from control: tools give the agent power, LangGraph gives control through explicit state, nodes, and conditional routing. Production-grade Agentic AI is traceable workflows, not magic."
"I'm a backend Java engineer. With no prior exposure to AI, I got a good solid foundation and a clear direction. This program changed my thinking about how we should implement enterprise-level RAG and build production-ready agents."
"Before joining, I struggled with GenAI concepts. The cohort helped me bridge the gap between a Data Scientist role and a GenAI role. I would highly recommend it to anyone who wants to transition with a strong foundation."
From verified cohort feedback and public learner reflections · 9.6/10 average experience rating in our latest cohort feedback survey.
Our way of investing directly in the learners who join first.
This is a foundation-to-intermediate program. If you have some programming exposure and a serious interest in Agentic AI, you'll do well here. The two pre-bootcamp Python foundation classes get you ready for day one.
Some Python comfort helps. If you're already writing Python, you're set. If you're new to Python but have other programming exposure, the pre-bootcamp classes will get you where you need to be.
Yes. Two dedicated pre-bootcamp foundation classes cover Python setup, environments, JSON handling, error handling, API calls, SDK usage, Git and GitHub, and the developer workflow around Claude Code and Cursor.
Every single week. Six weekly deliverables plus one final portfolio project. Every artifact lives in your GitHub.
Yes — Week 4 is a dedicated LangGraph module covering graph-based workflows, state schemas, conditional routing, memory, and human-in-the-loop.
Yes — Week 5 covers CrewAI, AutoGen, Agno / Phidata, and framework comparison, including when multi-agent systems help and when they don't.
Yes — Week 2 introduces the Model Context Protocol at real developer depth: client / server, tools, resources, prompts, and where MCP fits in modern Agentic AI systems.
Yes. Claude Code, Cursor, and GitHub Copilot are treated as first-class developer tools throughout the bootcamp — from the pre-bootcamp classes onwards.
Yes — Week 6 covers AWS AI foundations + Amazon Bedrock, Azure AI foundations + Azure OpenAI, GCP AI foundations + Vertex AI, plus deployment awareness across FastAPI and Docker.
Yes — six weekly deliverables plus one final portfolio-style project you design and submit as homework. Four project archetypes to choose from with clear component requirements.
Yes — every session is recorded and available inside your Manifold AI Learning portal.
You leave with real fluency, six hands-on artifacts, and a final portfolio build. From there, you can take your work into internal builds, client projects, or portfolio-strengthening projects — or continue into the Agentic AI Enterprise Mastery Bootcamp — our advanced flagship program covered in the "Next Step After This Bootcamp" section above.
No. This is a learning program. Manifold AI Learning does not offer or imply any job, placement, hiring, salary, or income guarantee. Outcomes depend on your own execution, portfolio quality, and market circumstances.
Since access to live sessions, recordings, community, and learning materials is provided after enrollment, refunds are limited once your account is activated. Please review the program details before enrolling. For any questions, contact support before purchase.
The Launch Cohort starts on 16 August 2026, from 8:00 AM to 11:00 AM IST.
Launch Cohort pricing is open at ₹19,999 + GST · Regular pricing returns to ₹24,999 + GST.
EMI available at checkout — from ₹1,144/month (India)
Enrol early and start prep with the Python Foundations Module — included for enrolments before the first class on 16 Aug.
Enrol in the Launch Cohort →Not ready to enrol yet? Register for the free demo session first →