Max Halford

Blog Links Bio SQL Cycle touring Pub quiz
new 4 November 2026 2026 - 11 - 04 Indexing Wikipedia on turbopuffer
8 September 2026 09 - 08 Squill: my canvas for writing SQL
1 September 2026 09 - 01 A brief guide to French polling data
20 July 2026 07 - 20 What's next for River (and myself)
2 July 2026 07 - 02 Will agents replace scripts?
4 June 2026 06 - 04 My unvarnished guide to solution engineering
11 May 2026 05 - 11 Autonomous web scraping with Claude Code
12 March 2026 03 - 12 Lower your warehouse costs via DuckDB transpilation
6 February 2026 02 - 06 Text classification with Python 3.14's zstd module
2 February 2026 02 - 02 Solving Détrak with brute force
9 January 2026 01 - 09 Row level lineage at Carbonfact
12 October 2025 2025 - 10 - 12 Scraping Google Calendar events
24 August 2025 08 - 24 Warmshowers sparks joy
30 July 2025 07 - 30 Do LLMs identify fonts?
9 June 2025 06 - 09 Thoughts on DuckLake
6 May 2025 05 - 06 The total derivative of a metric tree
8 February 2025 02 - 08 Minimizing the runtime of a SQL DAG
2 January 2025 01 - 02 Hard data integration problems at Carbonfact
26 September 2024 2024 - 09 - 26 Introducing icanexplain @ PyData Paris 2024
27 August 2024 08 - 27 @daily_cache implementation in Python
9 June 2024 06 - 09 LCA software: exit the matrix
17 May 2024 05 - 17 Cutting up shoes to measure their footprint
4 April 2024 04 - 04 A training set for bike sharing forecasting
27 February 2024 02 - 27 Fast Poetry and pre-commit with GitHub Actions
14 December 2023 2023 - 12 - 14 Decomposing funnel metrics
1 December 2023 12 - 01 Efficient ELT refreshes
26 October 2023 10 - 26 Online machine learning on the road @ IDE+A, TH Köln
16 October 2023 10 - 16 Sh*t flows downhill, but not at Carbonfact
9 August 2023 08 - 09 Answering "Why did the KPI change?" using decomposition
25 June 2023 06 - 25 Measuring the carbon footprint of pizzas
3 June 2023 06 - 03 Graph components with DuckDB
11 May 2023 05 - 11 For analytics, don't use dynamic JSON keys
28 April 2023 04 - 28 Metric correctness doesn't matter, consistency does
7 March 2023 03 - 07 Online gradient descent written in SQL
15 February 2023 02 - 15 Using SymPy in Python doctests
22 January 2023 01 - 22 Online active learning in 80 lines of Python
17 January 2023 01 - 17 Are Airbnb guests less energy efficient than their host?
13 December 2022 2022 - 12 - 13 The future of River
20 November 2022 11 - 20 Parsing garment descriptions with GPT-3
25 September 2022 09 - 25 Dynamic on-screen TV keyboards
6 September 2022 09 - 06 NLP at Carbonfact: how would you do it?
24 August 2022 08 - 24 Matrix inverse mini-batch updates
28 June 2022 06 - 28 A rant against dbt ref
9 June 2022 06 - 09 First IRL meetup with the River developers
7 April 2022 04 - 07 Online machine learning with River @ GAIA
4 April 2022 04 - 04 Fuzzy regex matching in Python
6 March 2022 03 - 06 OCR spelling correction is hard
5 March 2022 03 - 05 Comic book panel segmentation
9 February 2022 02 - 09 Online machine learning in practice @ PyData PDX
6 January 2022 01 - 06 The online machine learning predict/fit switcheroo
24 December 2021 2021 - 12 - 24 Weighted sampling without replacement in pure Python
17 December 2021 12 - 17 Online machine learning in practice @ Applied AI
10 December 2021 12 - 10 Online machine learning in practice @ LVMH
11 November 2021 11 - 11 Web scraping, upside down
26 October 2021 10 - 26 One year at Alan
7 October 2021 10 - 07 Manipulating ephemeral data with git
10 September 2021 09 - 10 Dashboards and GROUPING SETS
19 August 2021 08 - 19 Homoglyphs: different characters that look identical
10 June 2021 06 - 10 Automated document processing at Alan
8 June 2021 06 - 08 Text classification by data compression
11 April 2021 04 - 11 Reducing the memory footprint of a scikit-learn text classifier
7 April 2021 04 - 07 An overview of dataset time travel
26 February 2021 02 - 26 The challenges of online machine learning in production @ Itaú Unibanco
22 January 2021 01 - 22 Quelle est l’empreinte écologique du Big Data? @ Toulouse Tech
21 January 2021 01 - 21 Organising a Kaggle InClass competition with a fairness metric
14 January 2021 01 - 14 Converting Amazon Textract tables to pandas DataFrames
6 January 2021 01 - 06 What my PhD was about
17 November 2020 2020 - 11 - 17 Computing cross-correlations in SQL
3 October 2020 10 - 03 Unsupervised text classification with word embeddings
20 September 2020 09 - 20 Focal loss implementation for LightGBM
17 August 2020 08 - 17 A few intermediate pandas tricks
10 June 2020 06 - 10 A brief introduction to online machine learning @ Hong Kong Machine Learning Meetup
7 June 2020 06 - 07 The correct way to evaluate online machine learning models
7 May 2020 05 - 07 Online machine learning with decision trees @ Toulouse AOC workgroup
4 May 2020 05 - 04 Server-sent events in Flask without extra dependencies
17 April 2020 04 - 17 I got plagiarized and Google didn't help
12 April 2020 04 - 12 Our solution to the IDAO 2020 qualifiers
31 March 2020 03 - 31 Speeding up scikit-learn for single predictions
26 March 2020 03 - 26 Machine learning for streaming data with creme
10 March 2020 03 - 10 Global explanation of machine learning with sensitivity analysis @ MASCOT-NUM
26 February 2020 02 - 26 Bayesian linear regression for practitioners
17 December 2019 2019 - 12 - 17 Under-sampling a dataset with desired ratios
29 October 2019 10 - 29 The benefits of online machine learning @ Quantmetry
23 October 2019 10 - 23 The benefits of online machine learning @ Element AI
16 September 2019 09 - 16 Finding fuzzy duplicates with pandas
13 July 2019 07 - 13 A smooth approach to putting machine learning into production
28 June 2019 06 - 28 The benefits of online machine learning @ Airbus Bizlab
28 May 2019 05 - 28 Machine learning incrémental: des concepts à la pratique @ Toulouse Data Science Meetup
21 May 2019 05 - 21 Skyline queries in Python
11 May 2019 05 - 11 Online machine learning with creme @ PyData Amsterdam
6 May 2019 05 - 06 SQL subquery enumeration
22 April 2019 04 - 22 An approach based on Bayesian networks for query selectivity estimation @ DASFAA
3 February 2019 02 - 03 Morellet crosses with JavaScript
5 December 2018 2018 - 12 - 05 Streaming groupbys in pandas for big datasets
13 October 2018 10 - 13 Target encoding done the right way
26 April 2018 04 - 26 Stella triangles with JavaScript
24 July 2017 2017 - 07 - 24 Unknown pleasures with JavaScript
19 June 2017 06 - 19 Subsampling a training set to match a test set - Part 1
1 June 2017 06 - 01 Docker for data science @ HelloFresh Berlin
20 March 2017 03 - 20 Halftoning with Go - Part 2
4 March 2017 03 - 04 Grid paintings à la Mondrian with JavaScript
26 January 2017 01 - 26 A short introduction and conclusion to the OpenBikes 2016 Challenge
9 January 2017 01 - 09 Challenge Big Data @ TSE
27 November 2016 2016 - 11 - 27 Halftoning with Go - Part 1
30 March 2016 03 - 30 Predire la disponibilité des Velib' @ Toulouse Data Science Meetup
25 March 2016 03 - 25 Recursive polygons with JavaScript
10 September 2015 2015 - 09 - 10 The Naïve Bayes classifier
2 August 2015 08 - 02 An introduction to genetic algorithms
14 July 2015 07 - 14 Setting up a droplet to host a Flask app
3 June 2015 06 - 03 Visualizing bike stations live data
Kaggle icon
mail