AI systems · Full-stack engineering

Matthew Topham

AI Systems Architect & Full-Stack Software Engineer

I build interactive storytelling systems, multimodal generation pipelines, and local-first agent applications. Active Marinara Engine developer with 15+ merged pull requests across React, TypeScript, Node.js, prompt orchestration, media pipelines, persistent game state, testing, and documentation.

Resume GitHub Marinara PRs
Open to opportunities: AI engineering, full-stack software, interactive media, and flexible roles compatible with completing my computer science degree.

Experience

Pasta Devs · Marinara Engine

AI Systems Architect & Open-Source Developer
Remote
2026–Present

Active developer on a local-first AI chat, roleplay, and game engine supporting multimodal generation, customizable agents, persistent worlds, and multiple model providers.

  • Authored 15+ merged pull requests spanning React/TypeScript UI, Node.js services, shared contracts, persistence, prompts, documentation, and regression tests.
  • Built the end-to-end Game Mode storyboard and scene-video workflow: narrative-to-keyframe planning, image/video generation, galleries, automatic rendering, responsive playback, and archived-session replay.
  • Designed provider-aware prompt architecture for still versus animated workflows, per-character NovelAI captions, keyframe and duration controls, campaign art styles, and editable prompt review.
  • Improved reliability with lint/type/build checks, regression suites, bundle-budget validation, defensive timeouts, request queues, manual browser QA, and user-facing technical documentation.
Selected proof: storyboard/video pipeline · multi-character prompting · anime pipeline · session replay · prompt review reliability

Selected projects

School of the Ancients

Creator & Full-Stack Developer

Voice-driven Socratic learning platform where users converse with AI historical mentors across web and VR experiences.

  • Implemented React/Vite interfaces, Node.js and Supabase services, authentication, RLS-backed Postgres, lesson and quest creation, CI/CD, real-time voice AI, and Unity XR prototypes.
View project organization →

AI Radar

Creator

Python and GitHub Actions pipeline that ingests RSS feeds, filters and summarizes AI research/news with LLMs, and publishes a concise daily brief.

View repository →

Additional engineering

Built modular trading automation with Python, real-time SignalR feeds, API-based order routing, Supabase logging, and model-assisted analysis; automated Linux/Cosmos infrastructure backups, monitoring, alerting, and failover workflows.