Limited seats leftReserve a seat →
Limited seats left

Internship-grade agentic AI project, in one month.

Build a multi-agent AI app on a real repo, reviewed by the engineer who owns it.

Internship alternativeApprenticeship opportunityYour semester project
Reserve a seat →
Talk to us first →Reserve your seat for ₹2,999. The balance is due before the first session.
// Why this cohort

No internship yet?
Do the work an intern does.

A working engineer explaining code to a student

Build a real AI agent

A multi-agent HR assistant that calls tools and takes actions, with guardrails and logs.

Prove it with a Git repo

Your own repo, with every commit and pull request you made, ready to share.

Show it in your sem and resume

Present it as your semester project, and talk through it in interviews.

// How it works

Same repo. Study it, learn it live, ship on it.

  1. 01
    Self-paced

    Study the repo

    Lessons, design diagrams and a step-by-step guide on the AltGAN platform.

  2. 02
    Twice a week

    Learn it live

    The engineer who built it explains the design decisions, and what broke.

  3. 03
    Reviewed

    Build and ship

    Pick up tickets on the same repo and get every pull request reviewed.

// The curriculum

Four weeks, one production agent.

The project

Multi Agentic HR Application

Multi-agent orchestrationTool callingHuman in the loopAudit trails
  1. Wk 1Python and Git fundamentals. Multi-agent orchestration, handoffs and tool calling against real services.
  2. Wk 2Tool design, retries and recovering from failures. Human approval gates, evaluation and cost caps.
  3. Wk 3Design review of your agent so far, then code reviews on your tickets.
  4. Wk 4Final code review, then a demo day where you explain your choices.
01Introduction to Agentic HR Assistant Project5 lessons
Project Goal: Intelligent HR AssistantDocs
Problem Statement: HR Query & Access ChallengesDocs
Core Technologies: RAG, SQL, LangGraph, GeminiDocs
Project Learning ObjectivesDocs
Rationale for Technology ChoicesDocs
02Introduction to LangGraph for Agentic Workflows10 lessons
LangGraph Core Concepts: Graphs, Nodes, and EdgesDocs
LangGraph State ManagementDocs
Benefits of LangGraph for Agentic WorkflowsDocs
Understanding LangGraph CheckpointersDocs
Human-in-the-Loop (HITL) with LangGraph InterruptsDocs
Implementing a ReAct Agent with LangGraphDocs
Integrating Tools with LangGraph AgentsDocs
Advanced State Management Strategies in LangGraphDocs
Debugging and Tracing LangGraph WorkflowsDocs
Integrating LLMs and Tools within LangGraphDocs
03Hybrid RAG Indexing & Retrieval8 lessons
Navigating the Policy Indexing PipelineVideo
Hybrid Strategy: Dense vs. Sparse Trade-offsVideo
Metadata-Driven Ingestion: The Chunk ContractVideo
Contex-Aware Chunking LogicVideo
Context-Aware Chunking LogicCode rundown
Retrieval FlowVideo
Retrieval Flow: Fusion and RankingCode rundown
Storage Locking & Concurrency GotchasVideo
04Relational HR Data Engineering5 lessons
Touring the HR Data SandboxVideo
Decision: Why SQLite for Agentic Demos?Video
Relational Integrity as LLM GuardrailsDesign diagram
The Build & Seed PipelineCode rundown
The SQLite Foreign Key GotchaVideo
05LangGraph Orchestration & State Management14 lessons
The HR Agent Graph TourVideo
Scratch Space vs. Long-term MemoryVideo
Solving the Subgraph State EchoVideo
The AgentState Schema ContractVideo
Mental Model of Graph defined in the projectVideo
Agent GraphVideo
Orchestrating the Execution FlowCode rundown
The Scope Check GatekeeperCode rundown
How ReAct retrieval agent is implementedVideo
react agentVideo
the react retrieval agentCode rundown
Response Synthesis & State FinalizationCode rundown
Pruning State to Prevent Token BloatVideo
Handling the Invisible Response ScarVideo
06Tool Engineering & Agentic Reasoning7 lessons
The ReAct Retrieval PlannerVideo
SQL Security: Multi-Layered GuardrailsVideo
Regex-Based Row-Level EnforcementVideo
State Injection via InjectedStateVideo
Constructing the Retrieval AgentCode rundown
Parallel Multi-Query RAG ToolsCode rundown
UX of Agentic Reasoning: Streaming StatusVideo
07Human-in-the-Loop (HITL) Workflows4 lessons
HITL in Agentic WorkflowsVideo
State Persistence with InMemorySaverVideo
Implementing the Interrupt ToolCode rundown
The Re-execution Idempotency TrapVideo
08Streaming WebSocket API & Server8 lessons
The Streaming Server Entry PointVideo
Session Persistence via Thread-IDsVideo
Deep-Event Capture with Subgraph StreamingVideo
The Real-time Event ProtocolVideo
The run_turn Event LoopCode rundown
Handling Interrupts and ResumptionCode rundown
Frontend Streaming & Reasoning UICode rundown
The Case of the Missing EventsVideo
// Coding agents

Use agents to build. Learn to own the result.

You learn both to use coding agents and to build applications that contain agents.

  1. 01
    Before generation

    Write a focused spec

    Behaviour, files, constraints and acceptance criteria.

  2. 02
    During implementation

    Guide a bounded change

    Give context, inspect the plan, keep changes reviewable.

  3. 03
    After generation

    Review and verify

    Read the diff, run tests, check real behaviour.

  4. 04
    In your submission

    Show your contribution

    Note what the agent helped with and what you changed.

// Who's teaching

Real working engineers.

Not trainers. Engineers shipping this kind of work right now.

Venu Madhav

Venu Madhav · Co-Founder, AltGAN · IIT Delhi

Mentor engineer — reviews, 1:1s and unblocking

Built the AltGAN learning platform.

Abhishek PG · HCLTech

Lead engineer · 8+ years — owns the project and its reviews

Builds multi-agent automation for enterprise clients at HCLTech.

You learn the way apprentices do: under the engineer who owns the system.
// From the cohorts

Learn from engineers who ship. Hear it from people who did.

Backend developer moving into AI
“I write backend services all day and I kept reading about agents without ever seeing one run. Walking a real HR assistant end to end showed me where the model actually sits in a system I already understand. That was the part no article had given me.”
Ritheesh CRitheesh CDeveloper · 2 yrsAgentic HR assistantADP
Backend and DevOps, now agents
“I have shipped backend and infrastructure for a few years, so the engineering was familiar. What I did not have was how an agent is actually scoped, where it is allowed to act and what stops it returning something it should not. The walkthrough spent real time on exactly that.”
Sahithi BSahithi BSDE, Ex-Flipkart and Harness, 3+ yrs experienceAgentic HR assistantM.S at Penn State, IIT Roorkee
Got the review I never get at work
“The review was the part I did not expect. I pushed what I thought was a clean change and got back two questions about cases I had never considered. That conversation taught me more than the lessons did.”
Rishik RRishik RSoftware EngineerEnterprise chatbot for unstructured dataBITS Pilani

Ship an agent before your first job asks you to.

1 month · 2 live sessions a week · code and design reviews

// Before you join

Questions students ask first.

Can I use this as my college semester project?

Yes. You build the project on your own repo, with the design and review history to present it. Check your college's guidelines for the submission format.

Is this an internship?

It is an internship alternative: internship-grade work on a real repo, reviewed by working engineers, in a one month cohort. For early grads, it works as an apprenticeship under the engineer who owns the system.

When do the sessions run?

Two live sessions a week for four weeks.

What does it cost?

₹12,499. Reserve your seat for ₹2,999; the balance is due before the first session.