Sampath Kumar M
Senior Developer Programs Engineer @ Google Cloud
Working on next-generation agentic technologies — including the Agent-to-Agent (A2A) protocol under the Linux Foundation's Agentic AI Foundation (AAIF), the Agent Development Kit (ADK), and Google Cloud Agent Platform products. I write about Gemini, Google Cloud infrastructure, and engineering autonomous multi-agent systems.
Blog · GitHub · LinkedIn · Medium · Dev.to · Bluesky · PyPI · Docker Hub
Live Agentic Demos
-
PC Build Arena — MCP vs A2A
Live sandbox where AI agents compare MCP tools and A2A shop-clerk delegation by racing to buy expiring PC parts (30–45 s TTL) and assemble a working PC.
-
Distributed Data Warehouse — 5 Isolated MCP Sources
Five isolated DuckDB data warehouses (
orders,stores,customers,order_items,products— 168k+ rows), each exposed via its own read-only streamable-HTTP MCP server. Enforces one table per query (500 rows/page, 100 ms latency) with 50 challenge questions and zero-load live telemetry.Live Dashboard & 50 Questions · AI Catalog (JSON) · GitHub & 500-Run Report
Blog, Cookbooks & Speaking
-
Google Cloud Gemini Cookbook
Step-by-step, production-grade tutorials for Gemini and Vertex AI applications.
-
Writing & Technical Guides
40+ essays and deep dives on A2A protocols, memory systems, Google Cloud Next, and developer practices.
-
Speaking & Events
Keynotes, technical workshops, and conference sessions (Cloud Next, Kaggle AI Days, AI Day Sweden).
-
For AI Agents & LLMs
Machine-readable indexes:
/llms.txtand/llms-full.txt.
Open Source Tools
-
tk
Git-native ticket tracking for humans and AI agents: dependency graphs, priorities, zero setup.
-
quillscore
Grades the quality of developer content: offline-first, with actionable feedback.