TL;DR
Get the latest gadgets delivered free with Prime
- Fast, free delivery on millions of items
- Prime Video, Amazon Music and more included
- Member-only deals all year
The Pragmatic Engineer published an interview with Maude Lemaire about her infrastructure work at Slack and GitHub, including how Slack tested for a customer with 500,000 users and why GitHub’s database sharding was constrained by scatter queries. The discussion also covers performance decisions and Lemaire’s approach to checking AI-generated work; it is an interview, not an announcement of a new product or company initiative.
The Pragmatic Engineer has published an interview with Maude Lemaire, a principal engineer whose work has included performance infrastructure at Slack and database scaling at GitHub. The discussion describes specific challenges behind supporting large software products, including a planned 500,000-user Slack customer and limits on GitHub’s database sharding.
At Slack, Lemaire led the backend performance infrastructure team during nearly eight years at the company, according to the interview’s introduction. For a prospective customer with 500,000 users, she needed to test whether Slack’s service could handle the expected load. Her initial plan used headless Chrome instances, each running a Slack session. The source estimates that approach would have cost $0.37 per instance for a five-hour session, making a full run prohibitively expensive.
Lemaire instead built Koi Pond, a load-testing tool that starts slimmed-down Slack clients called “Koi.” Written in Go, the clients establish WebSocket connections without running full browser sessions. The report says the alternative addressed the cost problem; it does not provide measured test results or say whether the prospective customer ultimately signed up.
The interview also describes a database constraint at GitHub. The company’s Rails application used Active Record, and the report says the homepage could trigger about 100 SQL queries across three database clusters. Queries without a shard key had to reach every shard, a pattern known as a scatter query. According to the source, this limited GitHub from adding more than 16 database shards at the time being discussed. Lemaire also talks about using AI tools while personally verifying their output on foundational engineering work.
Scaling Without Overloading Systems
The examples show that adding capacity is not simply a matter of adding servers or database shards. How requests are generated and routed can determine whether a system remains responsive as usage grows. Slack’s WebSocket load test had to reproduce a large number of connections affordably, while GitHub’s scatter queries made additional shards less useful for some database work.
For engineers and technology leaders, the account offers concrete illustrations of the trade-offs behind infrastructure choices. A tool that makes application code easier to write can obscure the database work it triggers; a realistic load test can also become expensive if it simulates users with full browser sessions. These are examples from Lemaire’s experience, not evidence that every large service faces identical constraints.
Products Worth Considering
Amazon Kindle Scribe (64GB) with Premium Pen, Metallic Jade
Bose QuietComfort Ultra Earbuds (2nd Gen) – Wireless Noise‑Cancelling Earbuds with Mic, Immersive Audio, USB‑C Charging, Up to 6 Hours Batte
LG UltraGear 32GX850A-B 32-inch 4K UHD Glossy OLED Gaming Monitor
As an affiliate, we earn on qualifying purchases.
Lemaire’s Work Across Two Platforms
Lemaire spent nearly eight years at Slack before joining GitHub in 2024, according to The Pragmatic Engineer. At GitHub, she worked on database scaling as the company’s Copilot product was gaining attention. The interview also identifies her as the author of Refactoring at Scale: Regaining Control of Your Codebase.
The source frames the conversation around lessons from operating popular software products: Slack’s growth in enterprise customers, approaches to load testing, the scale of GitHub’s Rails monolith, and Lemaire’s use of AI in engineering work. It also recounts an earlier performance project at Rent the Runway, where she said the homepage could take up to 10 seconds to load because of a high-resolution image. She argued that improving performance mattered even without a direct competitor pushing the company to do so.
“The best competition is competition with yourself, knowing that you can do better, knowing that you can drive a better experience.”
— Maude Lemaire, as quoted in The Pragmatic Engineer
Results and Timing Not Specified
The source does not state the interview’s publication date, give performance measurements from the Slack load test, or confirm the outcome of the prospective 500,000-user customer relationship. It also does not specify the period when GitHub’s 16-shard constraint applied or describe whether the company later changed its database architecture.
The reported figures and engineering examples are presented in an interview recap, rather than a separate technical paper with independently reported measurements. The source gives no quantified before-and-after results for the systems discussed. Its account should therefore be read as a description of Lemaire’s experience and the publication’s summary of the conversation.
The Interview Is the Latest Update
The source points readers to the full episode, transcript, and timestamps through The Pragmatic Engineer, with the episode available on YouTube, Apple, and Spotify. It does not announce a follow-up release, a new Slack or GitHub infrastructure change, or a scheduled milestone. Further details about the load test’s results or GitHub’s later sharding approach would require additional reporting or statements from the companies.
Key Questions
What is the news about Maude Lemaire?
The Pragmatic Engineer published an interview with Lemaire about her engineering work scaling Slack and GitHub, including performance testing, database constraints, and her use of AI tools.
How did Slack test for a customer with 500,000 users?
Lemaire built Koi Pond, which used lightweight Go clients called Koi to create WebSocket connections. The approach avoided the expense of running a full headless Chrome session for each simulated user.
Why did GitHub’s database sharding face a limit?
The interview says queries without a shard key had to run across all shards. These scatter queries could become slower as the number of shards increased, and the source says they constrained GitHub to no more than 16 shards at the time discussed.
What does the source say about Lemaire using AI?
The report says Lemaire uses AI tools but personally checks their output for foundational engineering work. It does not provide a detailed list of tools or a broader policy for AI use at Slack or GitHub.
Source: newsletter.pragmaticengineer.com
