Now open in offline mode (no live sessions) due to high demand — enroll anytime and earn the same certificate.
Mastering Agentic AI — From First Agent to Production Loop
A deep, practitioner-first curriculum covering the full agentic stack — model selection, tool design, protocols, memory, planning, observability, evals, multi-agent orchestration, computer use, and security.
Based on your location (India), you qualify for a Purchasing Power Parity discount.
Secure checkout via Razorpay
India checkout supports cards, UPI, net banking, and eligible EMI through Razorpay.
About this program
Most "agentic AI" content stops at a ReAct loop and a vector database. This cohort goes much further: it covers the complete practitioner stack needed to ship agents to production in 2026 — choosing the right reasoning engine and routing strategy, designing tools and interfaces that agents actually use correctly, the protocol stack agents use to talk to tools and each other (MCP, A2A, ACP/UCP, AGENTS.md), layered memory and context engineering, the planning patterns that separate demos from reliable systems, observability and evaluation as disciplines rather than dashboards, production multi-agent orchestration patterns from companies like Shopify and Anthropic, computer-use and multimodal agents, and the security model every agent builder needs given that agents are now the top attack vector. It runs in offline mode — no live sessions, enroll anytime, and work through all 12 modules and 88 lessons at your own pace, ending with the same verifiable certificate as our live cohorts.
Who is this for?
Software engineers, ML practitioners, and technical builders who want to go past chatbot demos and ReAct tutorials into the full stack required to design, observe, evaluate, and secure agents in production
What you'll actively build & learn
Reasoning Engine Selection
Choose and route reasoning engines, thinking budgets, and cost tradeoffs for agentic workloads.
Tool & Protocol Design
Design agent-ready tools and risk ratings, and wire agents together with MCP, A2A, ACP/UCP, and AGENTS.md.
Memory & Evaluation Systems
Build layered memory and context-engineering systems, and run evaluation science and observability as a discipline, not a dashboard.
Multi-Agent Security
Orchestrate multi-agent systems and defend them against prompt injection and supply-chain attacks.
Time Commitment & Schedule
Self-Paced Modules
Flexible
No live sessions — work through all 12 modules and labs whenever suits you, in any order you need.
Hands-On Labs
~36 hrs total
Every module ends with a lab: real agents, real tool suites, real protocol workflows, real eval pipelines, real red-team exercises.
Module-Based Syllabus
Each module is structured around three things: what you'll cover, what capability you'll walk away with, and the concrete deliverable that moves you toward a working system of your own. Work through them in any order, at any pace.
12 self-paced modules, 88 lessons, work through them in order or jump to what you need
A personal grasp of the full agentic stack — model selection, tools, protocols, memory, evals, orchestration, and security — applied to your own agent projects
Practitioner-first, hands-on lessons with labs in every module — no live sessions
The agentic paradigm
- Mental models, failure modes, and the four-generation lineage every builder needs.
A clear mental model for what agents actually are, the five anatomy layers, and when not to use one.
A minimal working agent built in 60 lines with no framework.
Choosing your reasoning engine
- Picking wrong here costs 10x in debugging; picking right unlocks everything else.
Confidence choosing and routing between models for agentic work, including thinking budgets and cost tradeoffs.
A three-model benchmark across tool-call accuracy, context adherence, and cost per task.
Tool design
- Bad tool schemas are the #1 source of agent failure in production.
The ability to write agent-ready tool descriptions, risk ratings, and recovery-aware budgeting.
A five-tool incident-response suite with schemas, risk tiers, and idempotency.
The agent ecosystem
- The infrastructure layer every practitioner must understand in 2026: OpenClaw, NemoClaw, Hermes, and the framework landscape.
A working decision matrix for choosing frameworks, plus an understanding of build-vs-buy tradeoffs and supply-chain risk in skill marketplaces.
A local NemoClaw agent deployed with verified data isolation.
The protocol stack
- MCP is tool access.
- A2A is agent coordination.
- ACP is commerce.
- You need all three.
Fluency wiring agents to tools and to each other across vendors using MCP, A2A, ACP/UCP, and AGENTS.md.
A cross-vendor workflow combining MCP tool access with A2A delegation between two agents.
Context and memory architecture
- From stateless prompt-response to layered memory systems that improve across sessions.
The ability to design working, episodic, semantic, and procedural memory layers, and to compact context without losing information.
A multi-session research agent with in-context, vector-store, and summary-compaction memory layers.
Planning and reasoning patterns
- ReAct, Plan-Act-Reflect, spec-driven development, and the spec-execute split.
The judgment to choose the right planning pattern for a task, and why vertical-slice planning ships faster than horizontal phasing.
The same complex task run through three planning patterns with trajectory and cost compared.
Observability
- Traces, spans, cost attribution, and the tools elite teams actually run.
An understanding of the five observability surfaces and how to detect behavioral drift before it becomes an incident.
An instrumented agent with OpenTelemetry-compatible traces and a CI cost gate.
Evaluation science
- pass@k vs.
- pass^k, trajectory graders, LLM-as-judge, Verifier's Law, CI gating.
The ability to build a real eval system rather than a one-off spreadsheet, and to gate deploys on regression.
A complete eval pipeline: golden dataset from traces, trajectory grader, calibrated LLM judge, and a CI gate.
Multi-agent orchestration
- Sequential, fan-out, supervisor/worker, debate, swarm, and the factory model.
Fluency designing multi-agent systems with maker-checker verification, spawn budgets, and recursion guards.
A four-agent incident-response factory with investigator fan-out, synthesis, and a human gate.
Computer use and multimodal agents
- Beyond text-in/text-out — agents that see screens, navigate UIs, hear speech, and process video.
An understanding of the perception-action loop for GUI agents and the security risks unique to computer use.
A browser agent that autonomously completes a multi-step web workflow and verifies success via DOM inspection.
Agentic security
- Prompt injection, the Lethal Trifecta, non-human identities, and MAESTRO threat modeling.
The ability to threat-model an agent system and apply defense-in-depth against injection, memory poisoning, and supply-chain attacks.
A red-team exercise against your own agent with a MAESTRO-guided defense-in-depth writeup.
The syllabus builds toward a final proof of work.
The weekly syllabus is designed to stack toward a capstone that demonstrates what you can actually build. By the end of the cohort, you are not just finishing modules. You are presenting a concrete output that ties the learning arc together.
View Alumni CapstonesIndustry-Grade Certification
Earn a credential that actually matters. Every certificate is tied to your Capstone Project repo, valid for life, and optimized for your professional technical profile.
View Certification TiersYour instructor
Anubhav Srivastava
Anubhav has spent the past two decades building machine learning and AI systems across startups, large enterprises, and high-scale consumer platforms. He has worked on patented AI technologies, authored books, and founded multiple ventures, and is currently building a deeptech startup focused on physical AI. Known for combining technical depth with practical thinking, he enjoys breaking down complex ideas into clear, accessible insights and is driven by a curiosity for how technology can solve real-world problems.
From our students
Engineers at different levels share what they built and what changed.
500+
Engineers trained
25+
Engineering leaders
40+
SaaS startups
50+
Alumni network
Alumni at
“The most technically rigorous program I've attended. No fluff — just pure deep-dives into transformer blocks and swarm logic. It's about understanding how LLMs actually work.”
Siddharth S.
Staff Engineer · Build Your Own LLM
“LangGraph and multi-agent orchestration was the missing link for our production pipeline. Essential for developers who need to move beyond single-prompt engineering.”
Elena R.
Senior AI Engineer · Agentic AI
“Direct access to instructors who are actually shipping AI products. The focus on evals-driven development is unique — we implemented their RAG evaluation approach across our entire startup.”
Arjun R.
Tech Lead · Claude Code
FAQ
Common questions about courses, formats, and what to expect.