Building AI with AI
GH Film Hub v2: Year Two of the Grading System
The grading math still holds. Year two is about the hub: two seasons on one landing page, a player-facing rubric, and a week file that grades and publishes itself.
25 days ago
Code & Impact
Hi, I'm Mitchell Dawkins, a Senior Software Engineer with 10+ years of experience building production software, including four years at Dropbox. I build full-stack applications, AI-powered systems, developer tools, and sports technology that solve real-world problems. This site is where I share what I'm building, the engineering decisions behind it, and lessons learned along the way from AI-native development and self-hosted infrastructure to football technology and helping run a nonprofit.
Latest Projects
Fasted
An offline-first biblical fasting companion PWA for a 190-day journey — daily plans, scripture, check-ins, journal, streaks, and optional Supabase groups with Web Push reminders.
ReactTypeScriptViteTailwind+3
Learn more →
Thinker
A microlearning PWA that replaces doomscrolling with a one-card feed of ideas, news, Gutenberg books, and free learning sites — plus catalog-grounded Ask via OpenAI or Ollama.
ReactTypeScriptVitePWA+3
Learn more →
Wordpress Solutions Cloud
A self-hosted VPS platform for deploying and managing a fleet of WordPress sites, each running as an isolated Docker stack with Git-driven deploys, centralized MainWP management, automated backups, and a FastAPI + React control plane.
FastAPIReactTypeScriptDocker+2
Learn more →
NBA Stat Spot
Full-stack NBA analytics platform with AI-powered prop bet predictions, ML models trained on historical player data, real-time ESPN integration, and a Monte Carlo simulation engine for edge finding.
FastAPIReactTypeScriptPython+3
Learn more →
GH Film Review Pipeline
Automated data pipeline that transforms raw CSV film logs into polished weekly and season-long football player evaluation reports — complete with a static site, PDF exports, and GitHub Actions CI/CD.
PythonPandasReportLabHTML+3
Learn more →
WNBA Stat Spot
Full-stack WNBA analytics platform with AI-powered prop predictions, a real-time betting scanner, Monte Carlo simulations, and a multi-service backend built across PHP/Laravel and TypeScript/SvelteKit.
LaravelSvelteKitPHPTypeScript+3
Learn more →
Power & Context
A fully self-hosted microservices stack that turns any article URL into an NPR-style two-host podcast episode using local LLMs and zero-shot voice cloning — no paid APIs required.
PythonFastAPIDockerOllama+2
Learn more →
Video Processing Platform
A multi-tenant platform for processing, transcribing, and analyzing video content using AI — with custom model training, RAG-powered search, and tenant-isolated Docker containers.
PythonFlaskDockerAWS+2
Learn more →Latest from the Blog
Low to No Cost Solutions
Tab Recorder: Quick Browser Demos to MP4
I built a Chrome extension that records the whole browser tab, muxes H.264 MP4 in the browser, and downloads a file I can send to clients or other devs — no OBS, no conversion step.
about 2 months ago
Low to No Cost Solutions
Fasted: A 190-Day Biblical Fasting Companion
I built Fasted — an offline-first PWA that walks a June–December biblical fasting journey day by day: phase instructions, scripture, check-ins, journal, streaks, and optional group sync.
2 months ago
About Me
I'm a Senior Full Stack Software Engineer focused on building practical software that solves real-world problems. With 10+ years of experience across full-stack development, AI systems, and cloud architecture, I'm always exploring better ways to build, automate, and create meaningful technology.
Machine Learning
Hugging Face, OpenAI, Ollama, RAG, Whisper, Scikit-learn
Web Development
React, Next.js, TypeScript, FastAPI, Laravel
Cloud & DevOps
AWS, Docker, CI/CD
Automation
n8n, GitHub Actions, Playwright