8-week intensive · Cohort-based · Hands-on

BUILD
AI
Agents.

Go from "I've used ChatGPT" to building production-ready AI agents that think, use tools, remember context, and work in teams — in 8 weeks.

LangChain LangGraph OpenAI Assistants API MCP Tools Google ADK Streamlit Python
Apply for next cohort → View curriculum ↓
24
Live sessions
48hrs
Total instruction
6+
Working agents built
1
Production capstone
0
Prior AI experience needed
→ ReAct Agents → LangGraph Workflows → Vector Memory → Tool Calling → RAG Pipelines → Multi-Agent Systems → OpenAI Assistants API → NL-to-SQL → Prompt Engineering → Chain-of-Thought → LangSmith Evals → Guardrails & Safety → MCP Protocol → Streamlit Deployment → ReAct Agents → LangGraph Workflows → Vector Memory → Tool Calling → RAG Pipelines → Multi-Agent Systems → OpenAI Assistants API → NL-to-SQL → Prompt Engineering → Chain-of-Thought → LangSmith Evals → Guardrails & Safety → MCP Protocol → Streamlit Deployment
Why this course

Not another
tutorial series.

Most AI courses teach you to use tools. This course teaches you to build them. Every session has a working output. The course ends with a production-deployed agent in your GitHub portfolio.

⚡
Build from day one
Every session ends with working code. No theory-only weeks. By Session 3 you have a multi-agent pipeline. By Session 9 you have a full LangChain research agent.
🎯
Industry-relevant stack
LangChain, LangGraph, OpenAI Assistants API, and MCP are the frameworks hiring managers ask about in 2026. You will use all of them in production context, not toy examples.
🏗️
Real capstone project
The Data Analyst AI Agent capstone is a fully deployable Streamlit application with six tools, LangGraph orchestration, and a live demo. Ship it to your portfolio, not just your coursework folder.
🧠
Deep, not wide
We go past "call the API and hope." You will understand why the stop token prevents hallucinated observations, how the Google ADK orchestration model compares to LangGraph, and what happens inside a LangGraph StateGraph.
🔌
MCP tools integrated
Model Context Protocol is Anthropic's open standard for connecting agents to any tool. You will build and deploy your own MCP server — a skill almost no freshers have but every AI team needs.
🛡️
Production thinking
Evaluation, guardrails, prompt injection defences, cost analysis, and Docker deployment. The course doesn't end at "it works." It ends at "it's ready to ship."
Curriculum

24 sessions.
6 phases.

Three sessions per week on alternating days. One hour of focused theory, two hours of hands-on building. Every session produces a working artefact.

Wk 1 · Sessions 1–3 Foundations & Core Concepts
Bots vs Assistants vs Agents Multi-agent architecture LLM internals + Prompt engineering
Wk 2 · Sessions 4–6 Tools, Memory & Planning
Tool calling + parallel calls Google ADK ecosystem overview CoT + Tree-of-Thought planning
Wk 3 · Sessions 7–9 LangChain Deep Dive
LCEL + Runnables RAG end-to-end pipeline AgentExecutor + custom tools
Wk 4–5 · Sessions 10–15 LangGraph & Assistants API
StateGraph + nodes + edges HITL + conditional routing Supervisor–worker pattern Threads + Runs + Code Interpreter LangGraph hybrid patterns
Wk 6–7 · Sessions 16–21 DA Agent Capstone
DA workflow mapping NL-to-SQL + error recovery 6-tool architecture sprint EDA pipeline + LLM narrative Streamlit UI + peer demo
Wk 8 · Sessions 22–24 Production & Final Demo
LangSmith evaluation Guardrails + prompt injection Google ADK deployment Final presentations
What you'll build

Code that runs.
Agents that work.

Every assignment is a real system, not a textbook exercise. You push to GitHub, you demo it live, you own it.

Capstone Project
Data Analyst AI Agent
A fully deployable Streamlit application where you upload any dataset and an AI agent automatically profiles the data, detects anomalies, generates charts, writes a business narrative, and produces a downloadable report — all through a chat interface. Six specialised tools orchestrated by LangGraph, with OpenAI Assistants API as the code execution backend.
LangGraph OpenAI Assistants API Code Interpreter Streamlit Google ADK pandas LangSmith Guardrails AI
Week 1–2 Lab
ReAct Agent from scratch
Build the Thought → Action → Observation loop manually in Python — no framework. Understand exactly how the stop token prevents hallucinated observations and why this loop is the foundation of all agent frameworks.
OpenAI API Python JSON Schema
Week 3 Lab
RAG Research Agent
A LangChain agent with three tools: RAG over a custom document set, a Python REPL for live calculation, and Wikipedia search. Produces structured answers with source citations, traced in LangSmith.
LangChain LCEL Google ADK LangSmith
Week 4–5 Lab
Multi-Agent Content Pipeline
A LangGraph pipeline with Researcher, Writer, and Editor agents supervised by an orchestrator. Includes a human-in-the-loop approval gate that pauses execution, waits for review, and resumes — fully checkpointed.
LangGraph StateGraph MemorySaver HITL
MCP Integration
Custom MCP Server
Build and deploy your own Model Context Protocol server in Python — then connect it to a LangGraph agent as a tool node. Write the integration once; use it across every framework.
MCP Protocol stdio transport HTTP+SSE
The stack

Tools the industry
actually uses.

LangChain
Chains, agents, tools, memory — the backbone of most production agent systems
LangGraph
Stateful agent workflows with conditional routing, loops, and human-in-the-loop
OpenAI Assistants API
Threads, Code Interpreter, File Search — managed execution for complex tasks
MCP Protocol
Anthropic's open standard for connecting agents to any external tool or data source
Google ADK
Google's open-source Agent Development Kit — build, deploy, and orchestrate multi-agent AI systems with Gemini
LangSmith
Tracing, evaluation, and cost analysis for production agent systems
Streamlit
Ship your capstone agent as a web app — file upload, chat UI, chart display
Guardrails AI
Input/output validation and safety checks for production-ready agent deployment
Learning outcomes

What you'll be able
to do after.

✓
Build autonomous AI agents that use tools, manage memory, and recover from errors without human prompting at every step
✓
Design and implement multi-agent systems with orchestrator–worker topology, shared state, and parallel execution
✓
Build production RAG pipelines — from document ingestion through chunking, embedding, retrieval, and cited generation
✓
Implement stateful agent workflows in LangGraph with conditional routing, human approval gates, and checkpointed state
✓
Use the OpenAI Assistants API with Code Interpreter and File Search to automate complex data analysis tasks via natural language
✓
Write precise system prompts using the five-component framework — role, goal, knowledge, constraints, and output format
✓
Build and deploy a custom MCP server that exposes any Python tool to any compliant agent framework
✓
Evaluate, red-team, and add guardrails to production agents — including prompt injection defences and cost analysis
Who it's for

Built for people
just starting out.

You need Python basics and some curiosity. Everything else — frameworks, APIs, system design — is taught from scratch in context.

🎓
Final year students
  • CS, IT, or related engineering discipline
  • Know Python basics — loops, functions, classes
  • Some exposure to ML concepts or APIs
  • Want a portfolio project that stands out in campus placements
  • Prefer building over passive learning
💻
Recent graduates & freshers
  • 0–2 years experience, looking to specialise in AI
  • Comfortable with Python but new to LLM frameworks
  • Have heard of LangChain but never used it in production
  • Want something concrete to show in interviews
  • Aiming for roles: AI Engineer, ML Engineer, Prompt Engineer
🔄
Developers switching into AI
  • Background in web dev, data engineering, or backend
  • Strong Python — want to apply it to AI systems
  • Already built things but not with LLMs or agents
  • Want structured upskilling, not scattered YouTube tutorials
  • Need a real project to anchor their pivot story

Highly skilled trainers, very knowledgeable and supportive. It was a wonderful learning experience. Would highly recommend for anyone looking for internships or learning opportunities.

Abhishek Abhi

I had a great experience as a Full-Stack Intern, where I gained hands-on experience in full-stack development. I worked on real-time projects that challenged me and helped improve my skills. The teammates were very supportive and always ready to help, making the learning experience even more valuable.

Omprava Prusty

The training program covered a wide range of topics, starting from the basics and gradually progressing to more advanced concepts. I would like to express my gratitude to Raghu sir, who played a pivotal role in my learning journey. consistently demonstrated the qualities of an excellent mentor.

Shantha A H
Format & schedule

Structured.
Intensive. Real.

Session structure
Total duration8 weeks
Sessions per week3 sessions
Session daysMon · Wed · Fri
Session duration2 hours
Theory & Hands-on per session1 hour
Hands-on per session1 hour
Total sessions24 sessions
Prerequisites & delivery
Python levelBasics required
ML / AI prior experienceNot required
LangChain prior experienceNot required
Delivery modeIn-person / hybrid
AssignmentsWeekly + capstone
Capstone submissionGitHub + live demo
CertificateYes — on completion
Your instructor

Taught by someone
who builds this.

👨‍💻
AI Engineering Faculty
Course Designer · Practitioner

This course is designed and delivered by an AI engineering practitioner who has built production agent systems — not just demonstrated toy examples. The curriculum is grounded in real-world challenges: what actually breaks in production, what interviewers actually ask, and what architecture decisions actually matter at scale.

Every code example in the course was written and debugged before it was taught. Every concept was chosen because it shows up in real agent engineering work, not because it fills a syllabus slot.

LangChain · LangGraph OpenAI Assistants API Production AI Systems RAG Architecture MCP Protocol

Your agent
won't build itself.

8 weeks. 24 sessions. One production-ready agent in your portfolio. Applications open for the next cohort.

Apply for next cohort → Review curriculum first

SEATS ARE LIMITED PER COHORT · APPLICATIONS REVIEWED WEEKLY

Subhanu Technologies & Solutions © 2025 All Rights Reserved.