25+ years building stuff. Currently at Google as Group PM for Firebase - working on serverless products (Cloud Functions, Hosting, Storage, Extensions, App Hosting, Firebase ML) and our new AI framework, Genkit. Basically helping developers ship apps faster.

Led multiple products from zero to GA:

• Firebase App Hosting - Simplified serverless hosting for Next.js/Angular apps. No more config hell - just GitHub integration and automatic scaling.
• Firebase Genkit - Open-source AI framework for JS/TS developers. Making AI development as easy as Firebase made mobile backends.

Before Google: 14 years at Microsoft. Started as a dev consultant, ended up as Principal PM. Built Hot Reload for .NET/C++ (you're welcome!), shipped .NET MAUI 1.0, made XAML tools suck less, and helped rebrand VSTS to Azure DevOps. Also keynoted at .NET Conf and ran Build.

Started my career bouncing between banks, startups, and everything in between. Real-world experience taught me that technology can solve lots of problems, but the key is understanding your customers and what you're solving for them. Technology is fun, but building software is about the users.

Beyond tech: 6.5 years as NYPD Auxiliary Police in NYC, now doing C.E.R.T. with Florida fire department. Community matters.

I'm all-in on AI coding tools - Claude Code, Cursor, Windsurf, Gemini CLI, Firebase Studio, Jules. Here's what I've built with AI assistance:

Zero game dev experience → playable prototypes in hours. Nothing here is a fully working game, but it's a lot of fun exploring what's possible:

• Defend Ship - Tower defense with boats. Has weather effects!
• Maze Explorer - Touch-friendly maze game
• Space City - Build colonies in space
• Solar Explorer - Solar system sim with debug tools
• Explore Worlds - Side scroller with background ART pre-generated with Nano Banana
• Space Shooter - Defend your ship from enemies! Background art and enemies all generated by Nano Banana

As a PM, I build prototypes to explore concepts (not shipping these, but way better than just docs):

• Reimagined Firebase Hosting Docs - Experimental docs site with AI-assisted deployment onboarding
• Voice-Powered Firebase Studio (video) - Chrome extension that speaks code changes using AI summarization

• FishTankSimulator - 3D virtual aquarium with AI-generated fish. Upload photos to create custom fish that swim around. Three.js + Genkit magic!
• TextNarratorDemo - Text-to-speech with 30 voices and 24 languages. Built entirely with Firebase Studio prompts - includes auto-translation!
• learn-genkit-gemini - Using Gemini CLI to automate Genkit sample generation
• TextToSpeechGenkitDemo - Multi-speaker AI voice synthesis with Genkit
• Genkit-Ollama-VisionSample - Local AI text extraction from images
• Dmitry-CLI - a sample using Genkit framework + Ollama to talk to talk to local models

• Nano Banana MCP server - my very first MCP server + NPM package. I've built this to make it very easy to generated images with AI assistive tooling, with extra tools for image editing and optimization.

CUJ: I want a CLI that's always useful, whether I'm online or offline.

This is an early-stage exploration of a single command-line tool that adapts to your environment. The core idea is to provide a baseline of functionality even when you're offline, using a local model like Gemma3n. When you reconnect, it would seamlessly transition to using powerful, specialized agents powered by Gemini that can leverage external tools to solve complex problems. This is a proof of concept, and much work remains.

CUJ: I want to understand my codebase instantly.

This is currently a UX concept, exploring the idea of a CLI that can deeply inspect your project's code and explain any part of it on demand. The goal is a user experience where you could ask a question and be guided through the code with clear walkthroughs, helping you quickly understand complex logic or a new part of a project. While the core UX is the focus, this is still an early-stage project that doesn't yet function as intended.

CUJ: I want an intelligent system to manage complex, multi-step tasks.

This is a proof of concept for a task-based orchestrator that uses Gemini to break down complex requests into a series of manageable steps. The idea is that as a user, you would provide feedback, and the system would use Gemini to recursively refine and execute the workflow until the entire task is completed, ensuring every step is accounted for. The basics are working, but this is an early exploration with significant work remaining.

• ConfigureMacKeys - "Just in time" CLI for managing Mac env vars with auto-backups

• HelpMe - Single question, single answer command-line concept powered by Gemini or self-hosted models via Ollama. I started this as a proof-of-concept and now building an MVP with my friend, harris-boyce

• LinkedIn: dmitrylyalin
• Twitter: @lyalindotcom