This is a submission for the New Year, New You Portfolio Challenge Presented by Google AI
About Me
Hi everyone! I'm a Software Engineer based in Bangladesh with over 3 years of experience specializing in Java and Spring Boot.
Currently, I work at Technonext (a sister concern of US-Bangla Airlines), where I lead a small team building complex backend architectures for a ride-sharing application like Uber. While my "home base" is backend engineering with some Angular on the side, I believe a modern developer should never stop learning.
This challenge pushed me to dive into Google Cloud Platform for the first time — and I'm glad it did. Deploying to Cloud Run, configuring containers, and integrating Gemini AI opened up a whole new world.
But beyond the tech, I wanted my portfolio to have personality. Not just another skills grid and project list. Something that talks back.
Enter DS-7 — an AI assistant that knows my entire career and adapts its tone based on who's asking. Recruiters get professional answers. Developers get witty terminal responses.
This portfolio represents that journey — combining my backend roots, frontend skills, and the power of Google's AI tools.
Portfolio
💡 Pro tip: Click the pulsing terminal icon in the bottom-right corner. Try typing
ls,sudo, or just ask "Why should I hire him?"
(Note: The embedded site above is live on Google Cloud Run. Click the chat icon to ask the AI about my skills!)
How I Built It
I didn't want just a static HTML page. As a backend engineer, I wanted to showcase a full-stack architecture that is robust, scalable, and "smart."
1. The "AI-First" Development Process
This project was born in the cloud.
- Starting with Antigravity: I used Google Antigravity, Google's AI-first development environment, to generate the initial project structure. It instantly set up the boilerplate for a Spring Boot backend and Angular frontend, saving me hours of setup time.
- Refining with Gemini: Once the foundation was laid, Gemini served as my primary co-pilot. I relied heavily on it to "furnish" the project—refining the TypeScript logic, debugging Docker networking issues, and optimizing the Spring Security configurations.
2. The Tech Stack
- Frontend: Angular (built as a responsive SPA).
- Backend: Spring Boot (Java 17).
- Containerization: Docker (multi-stage builds).
-
Cloud: Google Cloud Run (Serverless deployment in the
asia-south1region). - CI/CD: Google Cloud Build (configured to auto-deploy from GitHub).
3. Google AI Integration
Beyond development, AI is at the core of the user experience:
- Gemini API: I integrated the Gemini API to power a chatbot that answers questions about my work history and projects. Recruiters can simply ask, "Does he know Spring Boot?" instead of reading a resume.
-
DevOps Automation: I even used Gemini to generate the specific
gclouddeployment scripts and set up the CI/CD pipeline, ensuring a smooth path to production.
What I'm Most Proud Of
I am most proud of the seamless integration of new AI tools into a professional DevOps workflow.
Taking a project generated by Antigravity, polishing it with Gemini, and having it deploy automatically to Cloud Run via a CI/CD pipeline felt like the future of software engineering. Seeing the "Hello World" turn into a fully functional, AI-powered application running on Google's infrastructure was the perfect start to 2026!
Top comments (4)
Hey Tushar! Just tested DS-7 and I'm seriously impressed. The guardrails are rock-solid.
I initially assumed it was RAG-based, but the responses are too tight. I'm building a similar agent for my AI/Data Science portfolio and would love to know:
Is this pure Gemini API prompt engineering with Spring Boot validation layers, or did you use a different approach?
Any high-level insights on how you structured the system prompt/guardrails would be amazing. Thanks for sharing! 🚀
Hey! Happy to answer your question. It's pure Gemini API Prompt engineering without any RAG involved, but there are two api calls, one for a validation, another for the real result. The validation only declares if your prompt will go to the next api call. This is expensive in a production environment. As my portfolio do not have many visitor and the cost is very low for Cloud Run, I let this solution stay.
I hope I answered your question. Feel free to ask if you have any more queries.
Hi Dabobroto! I just sent a connection request. I’m a fresher building my own AI portfolio agent and aiming for AI Engineer roles. I’m genuinely inspired by how you built DS-7. I’d love to learn from your experience—the real roadblocks you faced from scratch and how you solved them. Let's connect and chat there! 🙌
Sure