Agentic AI
Details
Welcome to the April'26 edition of PyData Ireland. This meetup is in association with Quantum Black, AI by McKinsey.
Agenda for the day:
- 17:30 – Arrival
- 18:00 – Start of talks (Details below)
- Jitendra Gundaniya: Building Agentic GenAI Pipelines with Kedro Builder (Beginner-Friendly)
- Dmitry Sorokin: Let AI Agents Manage Your ML & Data Pipelines
- 19:30 – Networking
-------------------------------
Talk 1: Building Agentic GenAI Pipelines with Kedro Builder (Beginner Friendly)
Speaker Name and Designation: Jitendra Gundaniya, Senior Software Engineer at QuantumBlack, AI by McKinsey
Speaker LinkedIn: https://www.linkedin.com/in/jitu11
Talk Description: This talk shows how Kedro Builder, a visual drag-and-drop tool, can lower that barrier. We'll walk through building a simple agentic pipeline visually dragging LLM context nodes, connecting prompts and tools, and exporting a working Kedro project, all without writing boilerplate by hand. Think of it as a learning ramp: see the pipeline structure first, then dive into the code.
Key Takeaways:
- How a visual builder can make agentic pipeline patterns (LLM + prompts + tools → agent → output) more approachable for beginners
- What `llm_context_node` is and how it simplifies GenAI pipeline creation in Kedro
- A live walkthrough of building and exporting a GenAI pipeline using Kedro Builder
- Where the visual approach helps (pipeline scaffolding, boilerplate) and where you still need code (agent logic, output schemas)
-------------------------------
Talk 2: Let AI Agents Manage Your ML & Data Pipelines
Speaker Name and Designation: Dmitry Sorokin, Senior Machine Learning Engineer II at QuantumBlack, AI by McKinsey
Speaker LinkedIn: https://www.linkedin.com/in/dmd40in/
Talk Description: This talk explores how traditional data & ML pipeline frameworks can be wrapped with an MCP server, enabling AI agents (e.g. LangGraph, Claude Code) to connect and manage pipelines programmatically. We’ll look at how this approach combines the deterministic nature of production data & ML workflows with the flexibility and reasoning capabilities of modern AI agents.
Key Takeaways:
- How to expose data & ML pipelines through an MCP server
- How AI agents can orchestrate and manage existing workflows
- A practical pattern for combining deterministic pipelines with agentic flexibility
