Start with a typed boundary
Produce structured outputs and a tool-calling loop that handles errors and retries.
PROFESSIONAL CERTIFICATIONApplied AI.
Applied AI.
Built to work.
Applied AI Professional Certification Program
Design, build and evaluate reliable AI agents—with guided labs, faculty insight and a production-style application of your own.
THE AGENT ENGINEERING PATH01 — 03
01 / FOUNDATIONS
Start with the loop.ModelsTools
02 / AI SYSTEMS
Connect the pieces.RAGMemory
03 / LAUNCH
Make it reliable.EvalsGuardrails
One application. Built layer by layer.
4Instructor-led days
40+Hours of learning & building
1Production-style agent app
IIT Hyderabad
ONE PROGRAM. YOUR NEXT COHORT.
Choose your chapter.
ENROLLING NOW24 October – 14 November 2026
Saturday lectures, on-campus + online, with guided project development between sessions.
See the October scheduleWHAT CHANGES FOR YOU
From concepts to capability.
01
Build reliable AI agents
Work with tool calls, state, structured outputs and failure recovery.
02
Connect knowledge and workflows
Engineer retrieval, memory and graph-based orchestration with human review.
03
Evaluate before you launch
Use tracing, quality checks, cost monitoring and guardrails to improve your application.
YOUR CONNECTED BUILD
One application. Built from the inside out.
Each module extends the same codebase. You leave with a complete application and the evidence behind it.
Add knowledge & memory
Ground the app in relevant documents and maintain useful state between interactions.
Coordinate the workflow
Add graph orchestration, enterprise tools and explicit human review points.
Measure how it behaves
Trace execution, evaluate quality and track failure, latency and cost.
Demonstrate the complete build
Submit source code and evaluation artifacts, present your solution and incorporate feedback.
Tools and patterns you’ll work with
PythonLangChainLangGraphQdrantLangFuseModel Context ProtocolFastAPIEvaluation & guardrails
THE LEARNING BLUEPRINTFour core modules.
Four core modules.
One connected build.
From the first tool call to a system you can evaluate, observe and improve.
01MODULE 01Foundations of reliable AI agents
Foundations of reliable AI agents
Model selection, structured outputs, tool calling, retry logic and recovery. Build a typed extraction workflow and tool-calling loop.
Lectures
- How agentic systems differ from simple chatbots
- Model selection, context windows, tokens, sessions, and state
- Structured outputs with Pydantic and schema validation
- Native tool calling, retries, rate limits, callbacks, and failure recovery
Hands-on labs
- Build a raw Python LLM client with session state
- Convert messy business emails into typed JSON objects
- Implement a tool-calling loop with exception handling and retry logic
02MODULE 02Knowledge, memory & production RAG
Knowledge, memory & production RAG
Embeddings, vector stores, chunking, reranking and memory. Build and evaluate a document-grounded assistant.
Lectures
- Semantic search, embeddings, vector spaces, and similarity
- Chunking, metadata filters, reranking, and query expansion
- Short-term memory, long-term memory, and session continuity
- Prompt injection risk in RAG and defensive retrieval patterns
Hands-on labs
- Spin up Qdrant or Chroma locally and run metadata-filtered queries
- Build a RAG assistant over policy and operations documents
- Add memory summaries and evaluate retrieval precision
03MODULE 03Graph-based multi-agent systems
Graph-based multi-agent systems
Orchestration, checkpoints, MCP, plugins and human review. Build a researcher–writer–reviewer workflow.
Lectures
- When to use one agent versus many agents
- Supervisor, choreography, planner-executor, reflection, and review patterns
- LangGraph state, edges, checkpoints, breakpoints, and resumability
- MCP, plugins, skills, and safe enterprise tool integration
Hands-on labs
- Build a researcher-writer-reviewer workflow in LangGraph
- Add checkpointing and resume a paused workflow
- Expose a local business API or file tool through an MCP-style interface
04MODULE 04Evaluation, observability & launch
Evaluation, observability & launch
Component tests, tracing, cost and latency, guardrails and deployment. Complete your application and evaluation harness.
Lectures
- Why agent evaluation differs from model evaluation
- Component tests for tools, retrieval, memory, planning, and final answers
- LangFuse tracing, cost tracking, latency, and failure analysis
- Guardrails, human-in-the-loop approvals, monitoring, and improvement loops
Hands-on labs
- Instrument an agent workflow with LangFuse traces
- Create evaluation cases for hallucination, retrieval misses, and tool failures
- Package the final app with FastAPI and a lightweight deployment checklist
MAKE SPACE FOR YOUR NEXT STEP
ENROLLING NOWThe October cohort, at a glance.
24 Oct 2026Foundations of reliable AI agentsCampus + online
25–30 OctGuided project developmentOnline support
31 Oct 2026Knowledge, memory & production RAGCampus + online
01-06 NovGuided project developmentOnline support
07 Nov 2026Graph-based multi-agent systemsCampus + online
08–13 NovGuided project developmentOnline support
14 Nov 2026Evaluation, observability & launchCampus + online
Final program phaseDemonstrations, feedback & assessmentDetails from the program team
The schedule follows the cohort selected above. The program keeps one permanent page across editions.
ACADEMIC DEPTH. FIELD EXPERIENCE.
Meet your instructors.
BEYOND THE SLIDES
A room full of possibility.
Questions, conversations and the
quiet satisfaction of making something work.
THE LEARNING STAYS WITH YOUReal people.
Real people.
A new way of thinking.
Reflections from the people
who have learned and built with us.
APPLIED AI · AUGUST 2026Excerpts from participant posts featured on the Applied AI program website. These experiences relate to Applied AI, not the upcoming FDE program.
In their own wordsParticipant video stories · Coming soon
VIDEO SERIES
MARK THE WORK YOU HAVE DONE
Earn your achievement.
Earn an IIT Hyderabad Professional Certificate after successfully completing the program requirements and assessments.
- A production-style AI agent application
- Source code and evaluation artifacts
- Assessment and demonstration feedback
OFFICIAL REGISTRATION & ENQUIRIES
Choose your cohort. Take the next step.
Registration for the October cohort is open. Use the registration form or contact the team with questions.
Open October registration form Enquire on this page Google Forms opens in a new tab. Form availability is managed by the program team.
YOUR NEXT CHAPTER
Build something worth deploying.
For software engineers, full-stack developers, data scientists and AI enthusiasts who have working Python experience.
Before you begin
Working Python knowledge, familiarity with REST APIs, basic LLM and generative AI concepts, and experience with OpenAI-compatible APIs. Prior LangChain experience is not required.
Groups & campus stays
Corporate group discounts may be available. Guest House accommodation at IIT Hyderabad can be requested at an additional charge, subject to availability. Contact the team for current tariffs and booking assistance.
OCTOBER COHORT INFORMATION
Early bird offer: ₹45,000 + taxesLimited seats
Standard fee
₹50,000 + taxes
Early-bird price applies to the first 20 registrations.
Runs from 24 October to 14 November 2026.
- All sessions and guided labs (hybrid: on campus + online)
- IIT Hyderabad certificate
- −Accommodation not included
Can I join the October cohort?
The October cohort runs from 24 October to 14 November 2026. Contact the program team for registration assistance.
Is the certificate automatic?
No. The IIT Hyderabad Professional Certificate is awarded after successful completion of the program requirements and assessments.
Do I need LangChain experience?
No. Working Python, REST API familiarity, a basic understanding of LLMs and experience using OpenAI-compatible APIs are the stated prerequisites.
Is accommodation available?
Accommodation at the IIT Hyderabad Guest House may be available for an additional charge. Contact the program team for availability and booking arrangements.