Earl Grey - DEV Community

Civic technologist building AI tools that help systems explain themselves. Clew Suite builder, NACM 2026 speaker. I write about hackathons, product judgment, agent readiness, and learning in public.
Community curators have picked four of your posts as their favorite gems!
You're a consistent community enthusiast! Keep up the good work by posting at least 2 comments per week for 24 straight weeks. The next badge you'll earn is the coveted 32!
Awarded for having a post featured in the weekly "must-reads" list. 🙌
One of Hacktoberfest 2026's top participants
Awarded for completing a Hacktoberfest 2026 Challenge. Thank you for participating! 🤝
Awarded for completing a Weekend Challenge. Thank you for participating! 💻
Awarded for submitting a valid entry to DEV's Big Summer Bug Smash powered by Sentry. Thank you for participating! 🐛
You're a dedicated community champion! Keep up the great work by posting at least 2 comments per week for 16 straight weeks. The prized 24-week badge is within reach!
Community curators have picked two of your posts as their favorite gems!
Awarded for completing at least one prompt in a Frontend Challenge. Thank you for participating! 💖
Awarded for completing the Hermes Agent Challenge. Thank you for participating! 💻
Awarded for the best use of Google AI in the April Fools Challenge 🏆
Award for winning a prompt in the WeCoded 2026 Challenge 🏆
Consistency pays off! Be an active part of our community by posting at least 2 comments per week for 8 straight weeks. Earn the 16 Week Badge next.
Awarded for utilizing the Cloud Run embed in a DEV post.
Awarded for completing the "Built with Google Gemini: Writing Challenge." Thank you for participating! 💻
Keep contributing to discussions by posting at least 2 comments per week for 4 straight weeks. Unlock the 8 Week Badge next.
Awarded for completing the GitHub Copilot CLI Challenge. Thank you for showing us how you power up your terminal workflow! 💻
Awarded for completing the Algolia Agent Studio Challenge. Thank you for participating! 🤖
Keep the community conversation going! Post at least 2 comments for 2 straight weeks and unlock the 4 Week Badge.
Awarded for publishing a top Google AI post.
For actively engaging with the community by posting at least 2 comments in a single week.
Awarded for writing and sharing your first DEV post! Continue sharing your work to earn the 4 Week Writing Streak Badge.
AI education is overwhelming — thousands of courses, most behind paywalls, few personalized. Clew Directive cuts through it. 4 questions, a personalized learning path from verified free resources, a PDF you keep forever. No signup. No data stored. Multi-agent system (Scout + Navigator + Curator) on AWS serverless. Free, private, accessible to all.
AgentisLux shows you what retrieval agents (ChatGPT, Perplexity, AI overviews) see when they read your site's raw HTML. For your second audience. Live at agentislux.io.
The zine for last night's WNBA games. Every final gets an issue from each team's side, in two AI voices, with a share card, a 10-second audio recap, and a foldable print zine. Stats are counted by code; AI only writes the words, and a fact lock drops what doesn't match. AWS Lambda, Bedrock, Polly, S3, CloudFront. Unofficial fan project.
Model archaeology on Lemonade: a 2019 model (GPT-2 XL) vs a 2025 model (Qwen3) on one laptop. The smaller, newer one runs ~11x faster on the same CPU. My reproducible benchmark even caught my own first headline as a bad run. It wasn't the hardware, it was the models. Live dashboard. AMD Lemonade Challenge.
Skills/Languages What tools and languages are you most experienced with? Are you specialized or more of a generalist?
AI assisted full stack prototyping, AWS serverless architecture, DynamoDB data modeling, agent readiness, retrieval focused UX, and turning hackathon builds into usable civic AI products.
Currently hacking on
Working on Agentis Lux.io (Perseus Clew), and Madame Minou, with an emphasis on agent readiness, retrieval aware design, civic AI, and practical tools that make complex systems easier to inspect.
Civic AI, agent readiness, hackathon builds, public sector automation, explainable systems, and thoughtful conversations about how modern tools can make complex workflows easier to inspect and trust.
Debugging cloud cost mysteries with AI
Details subtle auth bugs and community PR fixes
Create an account to connect with Earl Grey. You can also sign in below to proceed if you already have an account.
Building in public makes reference asks effortless
CPU capped at 15% during heavy builds
Built entirely via chat on a low-end Chromebook
Even a 120/120 score requires manual proof