Project overview
title: "My Google Gen AI Academy APAC Cohort 1 Experience" description: "My experience building and deploying AI agents with Gemini, the Agent Development Kit, Model Context Protocol, and Google Cloud." date: "2026-08-23" author: "Urvil Patel" slug: "google-genai-apac-cohort-1" tags:
- Gemini
- Generative AI
- AI Agents
- ADK
- MCP
- Google Cloud
- Hackathon published: true
My Google Gen AI Academy APAC Cohort 1 Experience
I recently participated in Cohort 1 of the Google Gen AI Academy APAC—an incredible learning experience focused on building practical AI solutions using Google’s generative AI ecosystem.
Participants had to choose two out of three available challenges. I selected Track 1 and Track 2, both focused on building production-ready AI agents.
Track 1: Build and Deploy AI Agents
Track 1 focused on building and deploying intelligent AI agents using Google Gemini and the Agent Development Kit (ADK).
During this track, I explored:
- Building AI agents with Gemini
- Designing agentic workflows
- Implementing tool calling
- Managing agent instructions and execution
- Connecting agents with external services
- Deploying AI applications to the cloud
Track 2: Connect AI Agents Using MCP
Track 2 focused on securely connecting AI agents with real-world data, tools, APIs, and external systems using the Model Context Protocol (MCP).
This track helped me understand how AI agents can move beyond simple conversations and interact with practical systems through structured and secure integrations.
I explored concepts such as:
- MCP clients and servers
- Tool discovery and execution
- Secure access to external systems
- Connecting agents with real-world data
- Designing reusable AI tools
- Building context-aware agent workflows
Live Demo
After several experiments, debugging sessions, deployment failures, and late nights, I successfully deployed a working version of the project.
When Things Did Not Go as Planned
Then came the hackathon—but nothing ever goes exactly as planned in this unpredictable world.
The longer you live, the more you realize that the only things that truly exist in this reality are pain, suffering, and futility.
Software development can sometimes feel the same way.
It involves unexpected errors, failed deployments, broken integrations, infrastructure problems, and moments when you have to start again because something refuses to work.
Deployment alone usually takes me two or three days, often ending with me staring at logs at 4:00 AM.
However, every failed attempt teaches me something valuable about:
- System architecture
- Debugging
- Cloud infrastructure
- Production deployment
- Time management
- Resilience
- Technical execution
This time, I invested a significant quantity of hard work, while still focusing on achieving quality in execution.
The final result may not have matched every expectation, but the experience considerably improved my understanding of building and deploying real-world AI systems.
What I Learned
Through these challenges, I gained practical experience with:
- Google Gemini
- Agent Development Kit
- AI-agent architecture
- Model Context Protocol
- Agentic workflows
- Tool calling
- External API integrations
- Google Cloud deployment
- Cloud Run
- Debugging production systems
- Building under strict deadlines
More importantly, I learned that building an AI prototype is only one part of the journey.
Turning that prototype into a reliable, deployed, and usable product requires an entirely different level of patience and engineering discipline.
Preparing for Cohort 2
After 51 days, I will participate in Cohort 2 with my team.
This time, I understand where I struggled, what could have been executed better, and how we should prepare differently.
Our next attempt will be:
- More focused
- More disciplined
- Better planned
- Technically stronger
- More strategic in execution
If I play, I play to win.
Cohort 1 was about learning, experimenting, failing, deploying, and growing.
Cohort 2 will be about execution—and winning.
Gratitude
I am grateful to Google and Google Cloud Run for providing the cloud infrastructure and credits that made this experience possible.
I am also thankful to Hack2skill for organizing these AI-powered cohorts and hackathons and giving developers the opportunity to learn by building real-world solutions.
This was not merely a competition. It was an opportunity to experiment, face challenges, improve my skills, and become a better AI engineer.
The journey continues.