Scaling Slack And GitHub With Maude Lemaire
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

The Pragmatic Engineer published an interview with Maude Lemaire about scaling infrastructure at Slack and GitHub, drawing on her experience at both companies. The discussion covers Slack’s approach to large customer loads, GitHub’s database and Rails challenges, and Lemaire’s practice of checking AI-generated work, particularly on foundational systems.

The Pragmatic Engineer has published an interview with Maude Lemaire, a principal engineer who worked on performance and infrastructure at Slack and database scaling at GitHub. The episode describes how Slack prepared for a prospective customer with 500,000 users and how GitHub’s database access patterns complicated efforts to add shards, offering practical examples of how product growth can stress systems beneath the interface.

Before you orderOffer from Amazon

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
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

At Slack, Lemaire led the backend performance infrastructure team during nearly eight years at the company, according to The Pragmatic Engineer. One challenge discussed in the interview was how to test whether Slack could support a potential customer with 500,000 users. Lemaire initially considered using headless Chrome instances to simulate sessions. The report says each instance would have cost $0.37 for a five-hour session, making a test at that scale prohibitively expensive.

Instead, she built Koi Pond, a tool that used slimmed-down Slack clients written in Go to establish WebSocket connections. The interview also explains why Slack limits custom emoji reactions to 23: an emoji can be distributed to every member of a large channel, multiplying the network work as reactions accumulate. A related problem, described as an event flood, occurs when a user connected to many busy channels receives a large volume of events at once.

At GitHub, where Lemaire joined in 2024, she worked on database scaling amid the growth of Copilot. The source describes GitHub’s Rails monolith and its use of Active Record, an abstraction that can make database access easier for developers but obscure the load generated by application behavior. According to the report, loading GitHub’s home page could involve about 100 SQL queries across three database clusters. Scatter queries—queries without a sharding key that must reach every shard—also limited how far the system could scale: the account given in the interview says GitHub could not add more than 16 database shards under those conditions.

At a glance
reportWhen: Published; the supplied source does not…
The developmentThe Pragmatic Engineer published a podcast interview with Maude Lemaire about the infrastructure and performance challenges she encountered at Slack and GitHub.

How Growth Exposes Infrastructure Limits

The examples show that supporting more users is not only a matter of adding servers. A feature that seems small, such as an emoji reaction, can generate a large number of deliveries when it reaches every member of a large channel. Meanwhile, a user active across many channels can become a concentrated recipient of events. These patterns make load shape and fan-out important parts of performance planning, alongside total user counts.

The GitHub account points to a different scaling pressure: application abstractions can hide the cost of database operations. If routine page loads trigger many queries, and those queries must be sent across shards, simply adding more database partitions may not solve the problem. For engineers and organizations running large products, Lemaire’s examples underscore why teams need to measure real system behavior and understand what their tools do underneath the application code.

Products Worth Considering

Bose TV Speaker Compact Soundbar with Bluetooth and HDMI-ARC, Black

Bose TV Speaker Compact Soundbar with Bluetooth and HDMI-ARC, Black

As an affiliate, we earn on qualifying purchases.

Lemaire’s Work Across Three Platforms

The interview spans Lemaire’s experience across several software companies. The Pragmatic Engineer says she spent nearly eight years at Slack, including leading its backend performance infrastructure team, before joining GitHub to address database scaling. She later worked on performance at Cursor, including time to first token, the delay before a model begins generating a response.

The discussion also reaches back to an earlier lesson from her career. In 2015, while working at Rent the Runway, Lemaire encountered a homepage that could take up to 10 seconds to load. The report says she was told the performance issue was less urgent because the company had no competitors to worry about. She chose to work on improving it anyway, arguing that teams could measure themselves against the experience they wanted to provide rather than waiting for competitive pressure.

The source presents these points as highlights from a podcast conversation, not as a new technical benchmark or an announcement of a product change at Slack or GitHub. It also notes that Lemaire wrote Refactoring at Scale: Regaining Control of Your Codebase. The episode includes discussion of her use of AI tools, with a distinction between routine assistance and foundational engineering decisions.

“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’s account of her Rent the Runway experience

What the Interview Does Not Establish

The source is a summary of an interview and does not provide a full operational account of either company’s systems. It does not specify when the Slack load test took place, whether the prospective 500,000-user customer ultimately signed, or what results Koi Pond produced in the test. The reported 23-emoji limit and GitHub’s 16-shard constraint are presented as details from the discussion; the source does not describe whether either limit remains in place today.

The report also does not give a current measurement of Slack’s event-flood frequency or GitHub’s database performance after the work Lemaire discussed. The supplied material links to an episode transcript and timestamps, but does not include the full transcript. Readers should treat the figures as reported examples from the interview, not as current company-wide benchmarks.

The Full Episode and Its Technical Detail

The immediate next step for readers seeking more detail is the full Pragmatic Engineer episode and its transcript, which the source says are available through the publication’s page and podcast platforms. Those materials may provide more of Lemaire’s explanations and the surrounding discussion than the summary alone.

No further Slack or GitHub announcement is described in the source. Whether the companies have since changed their emoji handling, event delivery, database query patterns, or shard architecture is not stated. The episode’s broader takeaway is grounded in Lemaire’s experience: as systems grow, teams must test the loads they expect, trace the costs hidden by abstractions, and verify foundational changes carefully.

Key Questions

What is the news about Maude Lemaire?

The Pragmatic Engineer published a podcast interview with Maude Lemaire about her work scaling infrastructure and performance at Slack and GitHub.

How did Lemaire approach testing Slack for 500,000 users?

The report says she built Koi Pond, which used lightweight Go clients to open WebSocket connections instead of running a full headless Chrome instance for each simulated user.

Why did GitHub’s database architecture limit shard growth?

According to the interview summary, many queries lacked a sharding key and had to run across every shard. These scatter queries became more difficult as the number of shards increased, contributing to a stated limit of 16.

What does Lemaire say about using AI in engineering?

The report says she uses AI tools but personally verifies their output, especially for foundational work where errors could affect the stability of the systems built on top.

Source: newsletter.pragmaticengineer.com

You May Also Like

Exclusive: Meta lays out details of May 20 restructuring in internal document

Meta has disclosed internal plans for a major restructuring set for May 20, including layoffs and organizational changes, according to an internal document.

LCS 1st Seed FlyQuest Sweeps Brazil’s FURIA 3-0

FlyQuest, the top seed from North America, defeated FURIA from Brazil in a 3-0 sweep at MSI 2026, marking a significant win in the tournament.

Colonel Sanders' Iconic Portrayals: Actors' Impact

From Darrell Hammond to Ray Liotta, the actors who portrayed Colonel Sanders bring a fresh twist to the iconic figure, leaving audiences intrigued for more.

Government Orders GitHub To Remove Bluetooth-based Chat App Bitchat: Jack Dorsey

Authorities have instructed GitHub to remove the Bluetooth-based chat app Bitchat, citing security concerns. The move involves founder Jack Dorsey.