Fractional CTO · Porto, Portugal
Fractional CTO for AI-built products. From prototype to production.
Your product works in a demo. Then a customer hits a bug, a deploy breaks sign-up, or the team loses hours to manual work around it. I take AI-built prototypes and early products to production, and automate the repetitive operations around them. Fixed scope, tested, shipped.

Current CTO role
NowCharging schedules for commercial electric van fleets, so every van is ready for its shift without going over the depot's power limit.
I own the optimisation, the live vehicle data, and the infrastructure underneath.
A seven-day plan · illustration
--:--
Konetik
Acquired 2023Machine learning on real vehicle usage data that told companies which of their vehicles could go electric, automating manual consulting work so it could sell at scale. Quartix PLC acquired the company in 2023, and I led the technical handover.
BJJ Graph
LiveAn interactive map of Brazilian Jiu-Jitsu positions and the moves between them. The source code is public.
Not Investment Advice
LiveA parody investing blog written entirely by a language model, built in a couple of days as an experiment.
Selected writing
- Based AphorismsOct 2026 · 5-part series
- 12x Developer: Using the 1% Compound Rule to Grow 1100%Dec 2025 · 14 min
- Are you still diversifying using the Markowitz model? Welcome to the 21st century.Jul 2020 · 7 min
- How to Export your Jupyter Notebook to Medium in 3 easy stepsJul 2020 · 1 min
Fractional CTO and machine learning engineer, based in Porto.
I started out writing trading algorithms for clients, then did machine learning research at INESC TEC, mostly on time series, and published four papers. At Konetik I built the machine learning that told companies which of their vehicles could go electric; Quartix PLC acquired the company in 2023 and I led the technical handover. I built HowRisky.ai end to end on my own. I'm now CTO of a SaaS that plans charging for commercial electric van fleets. As a fractional CTO I take on a small number of engagements each quarter, so each client gets senior attention.
Have a prototype your customers are starting to rely on?
Send a link, a repo, or a few lines about what you need. I reply within two working days.
Explore services and pricingOn the bench
Paused- HowRisky.aiRetirement risk estimates using Monte Carlo simulation. Built end to end on my own.
- FutureGiftAn ETF position as a gift, held for a child until they turn 18.
Papers
INESC TEC- Benchmark of Encoders of Nominal Features for RegressionWorldCIST 2021 · peer-reviewed
- Estimating the Likelihood of Financial Behaviours Using Nearest NeighborsComputational Economics 2023 · peer-reviewed
- Hierarchical Qualitative ClusteringarXiv 2020 · preprint
- A Review on Oracle Issues in Machine LearningarXiv 2021 · preprint