👤 Background
I’ve spent the last eight years building AI systems—not the kind that generate memes, but the kind that make decisions under uncertainty, process massive datasets, and run in production without falling over.
Started in urban engineering, quickly realized I needed more logic and less art. Switched to computer science. Built and scaled a retail business to fund my education, then left it to focus entirely on what actually kept me awake at night: making machines learn.
🛤️ The Journey
2016-2017 • First Company — Co-founded MasterCom, tried game development, realized I was more interested in the systems than the games. Pivoted to UI development with Qt/QML. Shipped 13+ projects. Learned what I didn’t want to do.
2017-2018 • Financial Markets — Discovered Forex trading. Six months of pure research before touching a demo account. Built backtesting systems, learned signal processing, feature engineering, risk management. Realized simple strategies don’t work—markets are adversarial.
2018-2020 • Deep Learning — Dove into machine learning. Andrew Ng’s courses, then straight into PyTorch. Computer vision, NLP, time-series forecasting. Read papers as they dropped from arXiv. Built a dual RTX 2080 Ti rig. Implemented everything from scratch to understand how it actually works.
2020-2021 • Dideo (Video Platform) — First real ML role. Built data pipelines processing petabytes of visual data. Custom Numba kernels that beat C++ implementations. Synthetic data generation systems. Learned production ML isn’t just about accuracy—it’s about throughput, maintainability, and human-in-the-loop workflows.
2021-2022 • Deep RL — Deep dive into reinforcement learning. Traditional RL, then Deep RL. Built custom Gymnasium environments. Trained agents on financial markets. Spent six months failing to make stable models—learned more from failure than from any tutorial. Built scalable RL frameworks from scratch.
2022-2024 • MetaScape (Fintech) — Built news scraping pipelines, sentiment analysis systems, portfolio optimization with RL. First experience with production CI/CD, structured logging, and the reality that 80% of ML engineering is data engineering.
2024-Present • ML Engineer (Contract/Freelance) — Reformulating trading as multi-armed bandits instead of sequential RL. Building systems that scale. Shipping open-source tools. Looking for the next challenge.
🛠️ How I Work
I move between research and production fluently. Half my time goes to research - reading papers as they drop, implementing novel architectures, reformulating problems in new ways, running experiments that might fail. The other half goes to production engineering - type-safe pipelines, monitoring, systems that run reliably at scale. I’m equally comfortable shipping a production API and prototyping a research idea.
Theory and practice inform each other. I read papers to understand principles and find breakthrough ideas, then build systems to test whether they actually work. Most research claims don’t survive real data and production constraints. The gap between “works in the paper” and “works in production” is where I operate - and where the interesting problems live.
Optimization isn’t premature if you understand the bottleneck. Custom Numba kernels, memory-mapped arrays, vectorized operations - these aren’t tricks when you’re processing 100M+ rows. But I profile first. Measure, identify the real constraint, then optimize deliberately. Bad infrastructure kills good research; good engineering enables it.
The best code is maintainable code. Clear structure, explicit error handling, comprehensive tests. Whether it’s a research prototype or a production system, code gets read more than it gets written. Clever code becomes a liability. Boring, well-documented code is an asset.
🎯 What I’m Looking For
Research labs, quant firms, or ML-focused companies where I can:
- Drive research on hard problems - Deep RL, financial ML, large-scale systems, or adjacent domains where depth matters
- Ship production systems that users actually depend on
- Work with people who value both rigorous experimentation and reliable engineering
- Move fast - from “interesting paper” to “working implementation” to “deployed system” in days, not months
I’m looking for teams where “let’s try that paper” leads to real implementation quickly, and where production systems are built by people who understand the research deeply. Places where building systems and pushing research forward aren’t separate tracks - they’re the same job.
📬 Contact
