What you'll be able to do
- Choose the right model for a task by quality, latency, cost and data-privacy needs
- Build a RAG system with chunking, hybrid search, reranking and source citations
- Design tool-using and multi-agent workflows with LangGraph and MCP
- Write evaluation sets that catch regressions before a release
- Defend against prompt injection, data leakage and runaway agent actions
- Deploy an agent with tracing, cost budgets and human approval steps
Who this is for
- Software developers with 1 to 10 years of experience moving into AI engineering
- Backend and full stack engineers asked to add AI features at work
- Data engineers and analysts who know Python and want to build LLM applications
- Tech leads who need to judge agent architectures, risks and costs
Before you join
- Comfortable programming in Python (or JavaScript, with some Python practice before you start)
- Familiar with Git, REST APIs and JSON
- Working experience as a developer is recommended; strong final-year students can apply
BF
BIGXSTAR Faculty
Practising Software Engineers
Taught by engineers who build and ship software for a living. Every lab is reviewed the way a pull request is reviewed at work.
Curriculum
1LLM Foundations for Engineers1 Week
Key Takeaways:
Tokens, Context Windows and How LLMs Generate Text
Hosted Models vs Open-Weight Models
Calling LLM APIs from Python with Streaming
Structured Output with JSON Schemas
Choosing Models by Quality, Latency, Cost and Privacy
2Prompt & Context Engineering1 Week
Key Takeaways:
System Prompts, Examples and Instructions That Hold Up
Context Design and Long-Context Trade-Offs
Prompt Templates, Versioning and Testing
Function Calling and Tool Contracts
3Embeddings, Vector Databases & Retrieval2 Weeks
Key Takeaways:
Embeddings and Similarity Search
Document Parsing, Chunking and Metadata
pgvector, Qdrant and Chroma in Practice
Keyword, Semantic and Hybrid Search
Reranking and Query Rewriting
4Production RAG2 Weeks
Key Takeaways:
End-to-End RAG Architecture
Citations, Grounding and Answer Quality
Access Control and Multi-Tenant Knowledge Bases
Keeping Knowledge Fresh: Re-Indexing and Lineage
Caching, Latency and Cost Optimisation
5Tool-Using Agents with LangGraph & MCP2 Weeks
Key Takeaways:
The Agent Loop: Plan, Act, Observe
Designing Safe, Typed Tools
Building and Consuming MCP Servers
State, Memory and Checkpointing in LangGraph
Retries, Timeouts and Failure Recovery
6Multi-Agent Workflows & Human-in-the-Loop1 Week
Key Takeaways:
Router, Supervisor and Hand-Off Patterns
When Not to Use Multiple Agents
Approval Gates and Escalation to Humans
Limiting Agent Permissions
7Evaluation, Guardrails & Security1 Week
Key Takeaways:
Building Evaluation Datasets
Automated and LLM-as-Judge Evaluation
Prompt Injection and the OWASP Top 10 for LLM Applications
PII Handling and Data Protection (DPDP Act basics)
Output Validation and Guardrails
8AgentOps: Deploy, Observe & Control Cost1 Week
Key Takeaways:
Serving Agents with FastAPI and Docker
Tracing and Observability for LLM Apps
Cost Budgets, Rate Limits and Model Routing
Release Process: Evals in CI
9Capstone: Ship a Production Agent1 Week
Key Takeaways:
Scope a Real Business Use Case
Build, Evaluate and Deploy the Agent
Architecture and Risk Write-Up
Demo and Review with Faculty
🏆
Generative AI & AI Agents Engineering Certificate
Awarded after completing the graded labs and deploying a capstone agent that passes its evaluation suite.
Frequently Asked Questions
Is this generative AI course live or recorded?
Live. Classes run online on weekends with an instructor, and weekday labs are reviewed by mentors. You can join from anywhere in India.
Do I need a machine learning or maths background?
No. This is an engineering course about building reliable applications on top of existing models. You need to be comfortable writing Python and working with APIs; you do not need to train models from scratch.
Which tools and frameworks will I use?
Python, FastAPI, hosted LLM APIs and open-weight models, LangGraph, the Model Context Protocol (MCP), pgvector, Qdrant or Chroma for vector search, Docker, and tracing tools for observability.
Can I do this course while working full-time?
Yes. It is designed for working developers: live classes are on weekends and labs can be completed on weekday evenings. Plan for about 8 hours a week.
How is this different from the Generative AI & Agentic Systems Engineer Program?
This course is a focused 12-week track for developers who already work in software. The 9-month Generative AI & Agentic Systems Engineer Program goes much deeper, including fine-tuning, multimodal systems and AI governance, and suits learners making a full career switch.
What is the fee for the Generative AI & AI Agents course?
The course fee is ₹60,000 for 12 weeks. Contact our admissions team for the next batch dates and payment options.