Build a production-grade agentic system in 12 hours — RAG, knowledge graphs, agents, MCP, evaluation and production architecture. Not another chatbot course.
Most GenAI courses teach you how to call an LLM. This one teaches you how to engineer an AI system.
You will not build another chatbot. Over 12 hours you take one enterprise problem through the whole engineering lifecycle — business problem, AI use case, architecture, knowledge and data layer, retrieval, agents and tools, MCP integration, guardrails and human approval, evaluation, observability, security, deployment — and end with a production-grade system you can defend.
The capstone is an Enterprise Intelligence & Action Agent: it understands a natural-language business question, retrieves from documents, databases and a knowledge graph, selects tools, runs multi-step workflows, asks for human approval before sensitive actions, answers with citations, and evaluates its own output.
The goal is AI engineering judgement, not framework memorisation. Every module keeps asking the same question: why this architecture, and not the other one? That is what separates an AI developer from an AI engineer.
USERS
|
v
API / UI Layer
|
v
AI Orchestration
|
+--------------+--------------+
| | |
v v v
RAG Agent Guardrails
| | |
+--------------+--------------+
|
+--------+--------+
| | |
v v v
SQL Graph MCP
| | |
+--------+--------+
|
v
Enterprise Systems
+-------------------------------+
| Evaluation / Observability |
| Security / Governance |
| Cost / Performance |
+-------------------------------+
AI Product Engineer · Generative AI, AI Systems & Software Architecture
Dr Ganesh works on practical engineering problems involving generative AI, LLM applications, AI agents, RAG systems, knowledge graphs, software architecture, code intelligence and application modernisation.
His approach is simple: don't just learn AI tools — learn how AI systems are engineered. This workshop covers practical engineering concepts and general industry knowledge, and does not expose proprietary client or employer intellectual property.
Basic programming experience. Python is recommended but not mandatory — the architecture reasoning matters more than the syntax. This is not a first programming course.
No. The workshop is framework-aware but architecture-first: you will use agent frameworks, vector databases and MCP, but the exact framework is treated as replaceable. The architecture should survive a framework change.
Two working projects and one combined system: an Enterprise Knowledge Assistant (RAG + citations + structured retrieval), an Enterprise Tool-Using Agent (tools + state + MCP), and the Enterprise Intelligence & Action Agent that combines LLM, RAG, graph, SQL, agent, MCP, guardrails, evaluation and observability.
Complete source code and repository structure, architecture diagrams, the RAG and agent implementations, tool and MCP definitions, an evaluation dataset and framework, a security checklist, a production architecture template, an Architecture Decision Record template, workshop notes and a certificate of completion.
Because that is the part nobody teaches. A whole module is a deliberately broken RAG app you have to diagnose, and another is attacking your own agent — hallucinated tool calls, infinite loops, prompt injection, overreach — and then building the controls. Debugging an AI system is the skill that transfers.
Agents are one module of fifteen. The workshop covers what surrounds them in an enterprise: knowledge graphs, MCP boundaries, approval gates, evaluation, observability, cost, security and governance — the parts that decide whether a demo ever reaches production.
The final session is built around defending your architecture: why RAG and not fine-tuning, how the agent selects tools, how you prevent infinite loops, how you evaluated retrieval, how you control cost, what you would change at ten million documents or a two-second latency budget. You practise answering those with a system you actually built.
Yes. The recommended format is 2 days × 6 hours, and it is run in-house or on campus built around your stack and your data. Get in touch for scheduling and fees.
Delivered live online or in-house, scoped around your stack and one of your real problems. GST invoice from Vithupro Infotech Pvt. Ltd.; we can raise a proforma first if your procurement needs a PO.
2 days · 6 hrs/day · 12 hours live · Live Online · On request