We’re hiring! Here’s an FAQ post with all the typical questions that I receive.

Over the last few months, we’ve been hiring for a Machine Learning Scientist in my team. This is always an exciting time — I get to meet and talk to a ton of interesting candidates from around the world. 🙌

I often receive questions about the role, team, or company via email, LinkedIn messages, or during calls with candidates. …


What is a feature store; why & how did we build one at Monzo?

The corner of the internet that I read is awash with posts about feature stores: systems that aim to be “the interface between models and data.

This idea has been around for quite a while, but there are now an increasing number of companies that are building feature store platforms or products — the concept is becoming established in the machine learning system operations arena.

A few months ago, I built a feature store at Monzo; in light of that, I thought I’d share some of…


Detecting a trivial sound in a quiet environment using deep learning.

⚠️ Note: You can jump to the complete blog post on my github pages blog; the first section is below.

🏡 Hack day idea

I recently moved our washing machine out to the garage, which meant that I couldn’t hear it beep when it was finished. I would have some “false positive” trips out there (through the drizzly British rain!) when timers went off, only to find that the thing was still going. How horrendous.

It had also been a while since I wrote some code just for the sake of building something…


Looking at the minutiae of discovering when analytics SQL is completely broken (the hard way).

⚠️ Note: You can jump to the complete blog post on my github pages blog; the first section is below.

📊 Context

It’s not uncommon for Data Science workflows to have large chunks of SQL. Maybe you have a sequence of queries that you run every day to produce dashboards, or maybe you have a bunch of queries that spit out features that you feed into machine learning algorithms. …


Here are some thoughts on the recent discussions around NLP transformer models being too big to put into production, and a dive into how we have shipped them at Monzo using the HuggingFace library.

💬 Background

Over the last couple of years, there have been a ton of exciting developments in natural language processing (NLP). In case you haven’t been working in this area, here’s the crash course: the development of deep pre-trained language models has taken the field by storm. …


I recently gave a couple of conference presentations about how we are thinking about speed when developing machine learning systems at Monzo. This post covers some of the background to the points I was making in my talks, as well as what we’re doing in the Monzo machine learning team to speed up our own work.

Speed is not a word that is regularly associated with machine learning teams. …


At Monzo, we use data to help our customers find their own answers faster.

Our customer support team are always on hand if you need to chat with a human, but that isn’t the only way to find answers: the help screen now has over 400 pages that answer common questions, from activating your card to travelling abroad.

As we work to make sure we can keep giving you world-class support as we scale, we’ve been using data to make changes to the help screen so you can find your own answers faster.

Measuring how helpful we are

If you scroll to the bottom of any page in the help screen, you’ll see a button that says ‘I can’t…


…being systematic about how you work, not just what you work on.

Most of the blog posts that I read about tech show how thorough typical workflows are: we take a systematic approach when trying to solve problems in software development, data science, and product management. So why not be just as systematic when thinking about how we work within our team and company?

I recently had the opportunity to think about this in some time that I had between jobs. As part of this, I read a book called The First 90 Days. …


My notes from the PAPIs.io 2018 conference in London

Last week, I went to the PAPIs.io Europe 2018 conference, which was held in Canary Wharf Tower in London.

The conference describes itself as a “series of international conferences dedicated to real-world Machine Learning applications, and the innovations, techniques and tools that power them” (and, from what I gather, the name papis comes from “Predictive APIs”).

I went down on the Thursday, the day that was dedicated to “Industry and Startups,” and took some notes on what I saw. Here’s a quick summary!

The view from the 39th floor of Canary Wharf Tower.

ML infrastructure with Kubernetes, Dask, and Jupyter

The morning keynote was by Olivier Grisel, who is probably best known for his immense contributions to…


…to power features in the Skyscanner app ✈️

In January 2017 I left Skyscanner, where I was a Senior Data Scientist. Like I did with this post a couple of years ago, I thought I’d summarise my journey by collecting some thoughts and lessons learned from the projects that I was involved in — which all focused on building different machine learning features for the Skyscanner app.

This post does not have any breakthrough insights on specific algorithms or technologies. Instead, I focus on a few broader aspects of data science: definining methodologies, dealing with uncertainty, building pipelines, and fostering a culture of machine learning.

5. My first…

Neal Lathia

Data Science, etc.

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store