Agentic AI
Building autonomous LLM agents — tool use, planning, multi-agent systems, and the failure modes that don't show up in a demo.
6 Weeks · Advanced · Live Online
Building a demo agent takes an afternoon. Building one that doesn't quietly fail in production is the actual skill — that's what this course teaches.
About this course
An advanced, hands-on course in building agentic systems: tool-calling, planning loops, memory, multi-agent coordination, and the reliability/observability problems that only show up once an agent runs unattended.
What you'll build
A tool-using agent that recovers from failures, and a small multi-agent system with more than one coordinating role and basic tracing built in.
What's covered
- LLM agent architecture: planning loops and the ReAct pattern
- Tool calling and function schema design
- Memory: short-term context vs. long-term/vector memory
- Multi-agent coordination patterns: orchestrator, hierarchical, peer-to-peer
- Observability and tracing for agent runs
- Failure recovery: retries, fallbacks, human-in-the-loop escalation
- Evaluation strategies for agentic systems (not just accuracy)
- Cost and latency management for agent pipelines
Curriculum
- Week 1: Agent architecture: planning loops, the ReAct pattern, tool calling
- Week 2: Tool schemas and function calling in practice
- Week 3: Memory and context management: short-term vs. vector memory
- Week 4: Multi-agent coordination patterns: orchestrator, hierarchical, peer-to-peer
- Week 5: Observability, tracing, and failure recovery strategies
- Week 6: Capstone: multi-agent system with evaluation and cost/latency tuning
Projects you'll build
- Tool-using agent — An LLM agent that plans, calls external tools/APIs, and recovers from tool failures.
- Multi-agent coordination system — Two or more agents with distinct roles collaborating on a task, with basic observability built in.
Who this is for
Developers who already have ML/AI fundamentals (or have completed our AI & Data Science track) and want to specialize specifically in agentic systems.
Why this one's different
Failure modes get equal billing with the happy path — tool failures, infinite loops, and context overflow are treated as first-class material, not edge cases mentioned once.
Pricing
- STANDARD: ₹19,999
- PREMIUM: ₹29,999
- INTERNSHIP: ₹8,999
FAQs
Do I need the AI & Data Science course first?
You need solid Python and some LLM API experience at minimum. The full AI & Data Science track isn't required but is the recommended path in.
Which frameworks are covered?
We teach the underlying patterns directly against LLM APIs first, then show how those map onto common agent frameworks.