TJ Klint — Rate Limiting: The Art of Saying No.
Healthcare.gov's launch day proved that a system without rate limiting doesn't slow down: it disappears. We'll look at why rate limiting matters beyond just "protecting servers," walk through five Redis-backed algorithms with honest tradeoffs, and get into the distributed systems territory where things like clock skew, failover strategies, and race conditions start to quietly break your assumptions.
Faizan Khan — AI-Native Docs or Human-Native Workflows? How to Evaluate API Documentation Platforms.
API documentation platforms are starting to split into two very different models. Some are becoming AI-native: optimized for structured specs, automated pipelines, SDK generation, and machine-readable outputs. Others are stronger at human workflows: WYSIWYG editing, narrative docs, permissions, reviews, and collaboration across multiple teams. In this talk, he’ll compare the major API docs platforms through that lens and explain why the real decision is not just about reference quality, but about whether your organization needs automation-first delivery, human configurability, or both.


Christophe Sirois — Design Is Not Vibes: How Engineers Can Talk About Design Without Guessing.
For many engineers, design discussions feel uncomfortable. The feedback sounds vague: “make it cleaner,” “this feels off,” “the UX isn’t great.” Compared to code reviews, design conversations can feel subjective and hard to reason about. But good design isn’t magic and engineers can learn to discuss it pragmatically.
Mikaël Francoeur — SQLite: more powerful than you think.
Did you know that SQLite can do much more than just CRUD? Join me as we explore SQLite's extensibility and advanced features. We'll see how to perform vector search, interact with REST APIs, synchronize databases with each other, and even edit zip files — all in SQL!



François Levasseur shared how building AI systems with user-written integrations exposed a tricky challenge: How do you enforce contracts and detect breaking changes when users generate the code? Existing tools fell short — so he built Jex, a TypeScript library that type-checks JSON Schemas. He walked us through the type theory behind it, key TypeScript nuances, and the design decisions that shaped the library.
Frédéric Harper took a step back from the AI hype cycle and focused on what really matters for senior developers and leadership. Beyond vibe coding and flashy demos, he explored the practical wins, the risks, the blind spots — and how to responsibly lead teams in an AI-driven landscape.



Chris Grass delivered the talk Beautiful Types: Designing TypeScript Libraries That Feel Handcrafted.
He showed what “handcrafted” TypeScript types look like in practice: readable hover tooltips, autocomplete that surfaces the right options, and absurdly helpful error messages. The talk looked at libraries that nail this—including Zod, Prisma, and Effect—to show how their public types shape the developer experience. Attendees left with a practical checklist and a few ways to sanity-check a library’s type surface before publishing.
David Desmarais-Michaud delivered Yoke: Code-based Kubernetes resource management and the journey from client to server-side package management.
This talk explored the current state of Kubernetes IaC and package management, tracing the shift from older client-side approaches toward emerging server-side and orchestration patterns. It also examined the idea of defining infrastructure with actual code, reflecting a software engineer’s path into DevOps and platform engineering and the effort to bridge those two worlds.
The session included reflections on how software engineering and platform engineering diverged over time, along with the gradual return of software development practices into platform engineering. It concluded with demos of code-first Kubernetes package management and server-side package management.



Michael Masson, CTO at Botpress, delivered the talk Kubernetes the Right Way: A Platform Engineering Approach to K8s. He acknowledged Kubernetes’ power and complexity, and showed how platform engineering provided a smoother, more manageable path for teams to benefit from K8s without forcing everyone to master it.
Mark Savic, Cursor Ambassador for Montreal and CTO at Bucky AI, walked us through Mastering Vibes: Best Practices for Pair Programming. He addressed the gap between AI coding tool promises and the real frustrations developers faced, sharing proven patterns from the community to reduce friction, improve flow, and make AI-assisted development feel effortless. The session offered practical tools for all skill levels.
Fred Lavoie, Founder and President at Deck, discussed Making Impossible Integrations… Possible. He demonstrated how authentication agents and a unified API made it feasible to integrate with any data source—even those without APIs—while ensuring secure, scalable, and reliable access to credentialed data.



Sylvain Perron, CEO and Co-Founder of Botpress delivered a talk titled llmz: The TypeScript AI Engine, where we learned how to use AI reliably—making LLMs perform complex actions safely and predictably. He explored llmz, an AI execution engine that wrote, ran, and reasoned in TypeScript to power LLM-native apps and features.
Alexandre Bouchard, CEO and Co-Founder of Hookdeck, presented Webhooks at Scale: Best Practices, Lessons Learned. Drawing on Hookdeck’s experience processing over 100 billion webhooks, he highlighted key architecture patterns, considerations for event-driven applications, and the future of webhooks with event destinations and gateways.














