Built things. Broke things. Wrote it down.

Field notes on software, systems, and making AI do real work. Written from the seat, not the comfortable distance.

Latest

4 min read

Five Years as a Tech Executive: What I Actually Learned

Five lessons from five years of hiring, firing, building, and occasionally breaking things.

4 min read

Good practices that actually matter

The ones that quietly determine whether a codebase is healthy at year three.

4 min read

The non-linear path: MSc thesis to CTO

Not a career advice post. A map of how the dots actually connected.

4 min read

How we use AI to write and review code without losing ownership

The workflow. The guardrails. The moments it goes wrong.

Topics

AI & LLMs in production

22 posts

On building with LLMs where it counts, and where it breaks.

  1. How we use AI to write and review code without losing ownership
  2. What I'm watching in 2026: reasoning models, long-context, and the next shift
  3. The agentic engineering loop
See all 22 →

Production ML & computer vision

12 posts

Field reports from running models on real data, not benchmarks.

  1. Model evaluation in production: how we score our CV pipeline
  2. YOLO on physical infrastructure
  3. What three years of running CV in production actually teaches you
See all 12 →

Leading engineering teams

12 posts

Notes from the CTO seat, mostly on how small teams stay small.

  1. Five Years as a Tech Executive: What I Actually Learned
  2. Good practices that actually matter
  3. The non-linear path: MSc thesis to CTO
See all 12 →

Building a startup

20 posts

Founding, fundraising, scaling, and the decisions you only make once.

  1. 99.9% uptime with a six-person team
  2. What enterprise compliance actually requires from an engineering team
  3. Cutting operational costs 30% while growing system usage 50%
See all 20 →

Field notes & essays

19 posts

Year reviews, conversations with founders, and early-career lessons that still apply.

  1. Ten years of building with AI: 2015 to 2025
  2. 2023: the year the ceiling came off
  3. Tim Draper's 'Hero Mindset' - what it means after two years of watching it work
See all 19 →

By date

The full archive, if you prefer chronology over topic.

2026

  1. Five Years as a Tech Executive: What I Actually Learned
  2. Good practices that actually matter
  3. The non-linear path: MSc thesis to CTO
  4. How we use AI to write and review code without losing ownership
  5. What I'm watching in 2026: reasoning models, long-context, and the next shift

2025

  1. Ten years of building with AI: 2015 to 2025
  2. What scales without headcount
  3. The agentic engineering loop
  4. Why I don't have a separate AI team
  5. 99.9% uptime with a six-person team
  6. Model evaluation in production: how we score our CV pipeline
  7. Agentic workflows: the ones that work and the ones that blow up
  8. Building a small team that punches above its weight
  9. The Claude API in production: a working engineer's notes
  10. Hiring for Startups: Takeaways from Brian Chesky

2024

  1. What enterprise compliance actually requires from an engineering team
  2. From Head of Engineering to CTO in 13 months
  3. Cutting operational costs 30% while growing system usage 50%
  4. GCP + Cloud Run + FastAPI: the stack I keep reaching for
  5. The OKR system that got us to 97.6% completion
  6. Multi-agent systems: what actually works in production
  7. Growing from 2 to 30+ installations in twelve months
  8. YOLO on physical infrastructure
  9. What I found when I joined Wasteer's engineering team
  10. What three years of running CV in production actually teaches you
  11. Technical due diligence from both sides of the table

2023

  1. 2023: the year the ceiling came off
  2. AI-assisted sprint planning: the actual workflow
  3. Multi-modal models and what they mean for computer vision products
  4. Prompt engineering is engineering
  5. Benchmarks vs. production: why they tell you different things
  6. Why I switched from GPT-4 to Claude for internal tooling
  7. The compliance problem with LLMs
  8. AI-assisted code review: the system we built and what it actually catches
  9. LLMs as infrastructure, not features
  10. Six weeks of GPT-4 in our workflow: what changed and what didn't

2022

  1. ChatGPT launched. Here's what I actually think.
  2. Speed as a product feature
  3. European Patent EP 4 323 939 B1: what we built and why we protected it
  4. Selling AI to enterprise: what technical founders get wrong
  5. Building ML infrastructure for under EUR 100K
  6. Getting to 99.6% accuracy in production

2021

  1. Classification vs. detection vs. segmentation: choosing right the first time
  2. Building for millions of requests: the architecture decisions that held
  3. Computer vision in the logistics industry: the specific problems nobody writes about
  4. Raising EUR 6M: the technical founder's fundraising notes
  5. Why we chose YOLO for anti-counterfeiting
  6. Founding Countercheck: what the first 90 days actually look like

2020

  1. Tim Draper's 'Hero Mindset' - what it means after two years of watching it work
  2. 50 conversations before writing a line of code
  3. The EIR model: what it is, how to use it, and when to leave
  4. GPT-3 and the moment the ceiling got removed
  5. Moving a 500-person global program online in three weeks
  6. Anti-counterfeiting as a computer vision problem: early notes

2019

  1. The decade in AI: what actually changed between 2010 and 2019
  2. What Zillion Pitches taught me about building with AI
  3. Aaron Levie on building Box: the distribution insight nobody talks about
  4. Why most AI features in startup products are theater
  5. Tony Hsieh on culture: notes from a Draper session
  6. What Phil Libin talked about - and why product longevity is harder than product growth
  7. GPT-2 dropped. Here's what I think it means.
  8. Speech-to-text in production: latency, accuracy, and the tradeoffs nobody documents

2018

  1. NLP in 2018: mostly heuristics with a transformer on top
  2. What Biz Stone talked about - and why it stuck
  3. Draper University: what it looks like from the inside
  4. Sentiment analysis in production: the things the papers don't mention
  5. Building pitch analysis with AI: what the algorithm can and can't tell you about a founder
  6. IBM Watson: the gap between the demo and the integration
  7. San Francisco, 2018: What the City Taught Me About AI That No Conference Could

2017

  1. Deep Learning Became Boring - and That's the Most Important Thing That Happened This Year
  2. Why I'm Going to San Francisco
  3. One Year as a First-Time CTO: The Honest Retrospective
  4. Expanding to the US With No Network and Too Much Confidence
  5. The Gap Between "AI-Powered" and AI That Actually Works
  6. Azure Cognitive Services in 2017: An Honest Review From a Startup CTO

2016

  1. Shipping to the App Store and Google Play Simultaneously: What Breaks, What Holds
  2. What They Call You When You Do Everything: On Startup Titles and What They Actually Mean
  3. The Microsoft BizSpark Programme: What $120K in Azure Credits Actually Gets You
  4. Winning Y2FI and What Nobody Tells You About Startup Competitions
  5. Building a Social App Solo: Xamarin, iOS, Android, and a Backend, All at Once
  6. TensorFlow 0.x: First Impressions From Someone Who Had No Idea What a Computation Graph Was

2015

  1. The Year I Stopped Being Afraid of Machine Learning
  2. Why I Rewrote My MSc Thesis Project Three Times
  3. Understanding Backpropagation Without the Math Degree
  4. My First scikit-learn Classifier and Everything I Got Wrong