Hi, I'm
Tomas Gonzalez
I'm an Engineer
|
Skills
Core Strengths
Proficient In
Also Skilled In
About Me
I'm a Full-stack engineer who can build anything you need, especially with AI/ML. From facial recognition pipelines and retrieval-augmented generation to full web and mobile applications and the infrastructure that keeps them all running securely. I care about the full picture: not just getting a model to work, but making it fast, secure, and reliable enough to trust with real users. I lean heavily on AI-assisted development tools to ship more efficiently without cutting corners, and I've taken projects from idea to production myself, including designing and building a full community platform from scratch, solo. Outside of engineering, fitness is a huge part of who I am. Training combat sports, lifting, and running with the same mindset I bring to shipping code: stay focused, keep improving, work efficiently, and don't quit when it's hard. When I'm not at the gym or a keyboard, I'm usually deep trying to learn something new. Currently in NJ, and always looking for the next hard problem worth solving.
Experience
Founder & Solo Developer
Eleutheria
Dec 2025 – Jun 2026
- Designed, built, and shipped a full-stack anonymous community platform from scratch, solo live in production for real users
- Implemented layered security (rate limiting, parameterized queries, CSRF/XSS protection) and passed penetration testing across 8 attack categories
- Containerized with Docker Compose, deployed to DigitalOcean behind Traefik, with automated CI/CD via GitHub Actions
Full Stack Engineer
Cadooga · Remote
Jul 2025 – Apr 2026
- Owned the reverse image search system end-to-end — architecture, implementation, and deployment — using InsightFace/RetinaFace for embeddings and Qdrant for high-accuracy vector search at scale
- Built a public records search platform end-to-end, including automated data collection across all 50 U.S. states and a React Native mobile frontend
- Architected AWS cloud infrastructure and built a serverless ETL pipeline processing 128GB of data via AWS Glue
- Retained as sole engineer after a company-wide reduction in force; hired and mentored a junior engineer
Full-Stack / Cloud Engineer
Theoforge · Newark, NJ
Dec 2024 – May 2025
- Built a backend API from scratch using Python, FastAPI, and Jinja
- Wrote and maintained CI/CD pipeline configurations, reducing manual deployment overhead
- Reviewed junior developer code and provided feedback to maintain code quality.
Projects
Masi Memory
Masi Memory is a production RAG (retrieval-augmented generation) system built to mimic how human memory actually works. Memories aren't all equal, and how easily one resurfaces depends on how often it's actually mattered, not just how long ago it happened. A guardrail independently verifies every AI-generated answer against its source memories using embedding similarity before reinforcing them, preventing unsupported claims from strengthening the wrong memory.
AI Portfolio Chatbot
An AI assistant embedded in an earlier portfolio, answering questions about my background using retrieval-augmented generation. Weighed a self-hosted model against API-based inference on cost and latency before settling on Groq for fast, cheap generation, and chose a lightweight Lite-RAG retrieval approach over a full vector database. A vector DB would have been overkill for a dataset this size. Also benchmarked JSON against the newer TOON format for token efficiency, and found JSON came out slightly cheaper for this specific dataset.
Quake 4 Mod
A gameplay mod for Quake 4, built in C++ against the game's SDK. Overhauled weapon functionality, added a custom leveling/progression system with special ability upgrades, and wave-based enemy spawning.