How do you deploy federated front ends to the edge in 150 milliseconds? In this episode, Zack Chapple, CEO and Co-founder of Zephyr Cloud, and Nestor Lopez, Platform Engineer at Zephyr Cloud, break down everything developers need to know about micro frontends, module federation, and deploying at global scale without the infrastructure pain.

Zack's journey started at a consulting company working with enterprises like SAP to add module federation support to Angular, which eventually revealed all the pain points of scaling federated architectures. That led to Medusa, then to Zephyr Cloud, the platform he describes as "Kubernetes for the front end." Nestor's path started eight years ago with Sencha.js and iframes, long before module federation existed, and brought him to Zephyr through open source contributions to TRPC and other projects.

We cover why module federation is "Docker for the front end," how Zephyr deploys with one line of code and no CI/CD pipeline, their reverse tree shaking technique that recomposes federated bundles into a monolith at the edge, how Nestor deployed 5,200+ micro frontends as a single video, their federated MCP server for enterprise AI orchestration, and a TC39 proposal to fix ESM module unloading in V8. We also talk about pricing, open source contributions, and what it's really like to build a startup with four kids.

Whether you're an enterprise team trying to ship frontend independently across dozens of teams, or a solo developer who just wants to deploy without setting up a CI/CD pipeline, this conversation covers the full spectrum.

Key Topics:

- Micro frontends explained through the microservices and Kubernetes analogy

- Module federation as "Docker for the front end" and Zephyr as the orchestration layer

- End-to-end walkthrough: from bundler to global edge deploy in ~150ms

- No repo required, Zephyr hooks into any bundler and deploys on build

- Reverse tree shaking: monolith performance with micro frontend dev experience

- The Chrome extension for hot-swapping MFEs in any environment

- Federated MCP servers built on module federation for enterprise AI

- TC39 proposal to fix ESM module unloading and enable live HMR on Node.js

- Bring your own cloud: Cloudflare, AWS, Fastly

- Pricing: free for solo, $19/seat for teams, org-wide for enterprise

- Mobile support through Metro and desktop through Tauri

- Open source contributions and financially supporting projects like RSPack, SWC, and Tailwind

🔗 FOLLOW ZACK

💼 LinkedIn: https://www.linkedin.com/in/zackarychapple/

🐦 X/Twitter: https://x.com/Zackary_Chapple

🐙 GitHub: https://github.com/zackarychapple

🔗 FOLLOW NESTOR

💼 LinkedIn: https://www.linkedin.com/in/nstlopez/

🐦 X/Twitter: https://x.com/nstlopez

🌐 Blog: https://nstlopez.com

🎙️ FOLLOW & SUBSCRIBE

📸 Instagram: https://www.instagram.com/senorsatscale/

📸 Instagram: https://www.instagram.com/neciudev

🎙 Podcast URL: https://neciudan.dev/senors-at-scale

📬 Newsletter: https://neciudan.dev/subscribe

💼 LinkedIn: https://www.linkedin.com/in/neciudan

💼 LinkedIn: https://www.linkedin.com/company/senors-scale/

📚 ADDITIONAL RESOURCES

- Zephyr Cloud: https://zephyr-cloud.io

- Module Federation: https://module-federation.io

- RSPack: https://rspack.dev

- Hono: https://hono.dev

- shadcn/ui: https://ui.shadcn.com

#MicroFrontends #ModuleFederation #ZephyrCloud #Frontend #WebDev #PlatformEngineering #DevEx #EdgeComputing #Kubernetes #SenorsAtScale #OpenSource #Startup

💬 What's the most painful deployment workflow you've ever had to deal with? Share your stories in the comments!

William Morgan is the CEO of Buoyant and the creator of Linkerd, the world's first service mesh and a CNCF graduated project powering production Kubernetes infrastructure at thousands of companies. Before founding Buoyant, William spent nearly four years at Twitter as a software engineer and engineering manager, where he shipped core platform features like the Twitter photo service and embed timelines — and watched the legendary monolith-to-microservices transformation unfold firsthand.

In this episode, we cover what it was like engineering at Twitter during the fail whale era, how decomposing a monolith introduces entirely new networking challenges, why William invented the term "service mesh," and how Linkerd gives platform teams reliability, security, and observability without developers having to think about it.

Whether you're a platform engineer running Kubernetes in production, an SRE trying to make sense of service-to-service communication, or a developer curious about what infrastructure teams actually do — this conversation is packed with hard-won lessons from a decade of building critical open source infrastructure.

🔸 Key Topics:

- Engineering at Twitter in 2010: the Rails monolith, Scala rewrite, and microservices transformation

- How replacing function calls with network calls changes everything

- What a service mesh is and why the term had to be invented

- Control plane vs data plane architecture

- Why Linkerd rewrote its proxy from Scala/JVM to Rust

- Latency-aware load balancing, mTLS, and protocol detection

- Multi-cluster communication and mesh expansion to VMs

- Common service mesh implementation mistakes

- Linkerd vs Istio: William's honest take

- Open source sustainability and enterprise monetization

- The enterprise sales journey from engineer to CEO

- Book recommendations: Hyperion, Gideon the Ninth, The Book of the New Sun

🔗 FOLLOW WILLIAM

💼 LinkedIn: https://www.linkedin.com/in/wmorgan/

🐦 X/Twitter: https://x.com/wm

🌐 Buoyant: https://buoyant.io

🎙️ FOLLOW & SUBSCRIBE

📸 Instagram: https://www.instagram.com/senorsatscale/

📸 Instagram: https://www.instagram.com/neciudev

🎙 Podcast URL: https://neciudan.dev/senors-at-scale

📬 Newsletter: https://neciudan.dev/subscribe

💼 LinkedIn: https://www.linkedin.com/in/neciudan

💼 LinkedIn: https://www.linkedin.com/company/señors-scale/

📚 ADDITIONAL RESOURCES

- Linkerd: https://linkerd.io

- Buoyant: https://buoyant.io

- Linkerd Getting Started: https://docs.buoyant.io

- Linkerd GitHub (Proxy): https://github.com/linkerd/linkerd2-proxy

- Hyperion by Dan Simmons

- Gideon the Ninth by Tamsyn Muir

- The Book of the New Sun by Gene Wolfe

- Simon Willison's Blog (AI/LLMs): https://simonwillison.net

#Linkerd #ServiceMesh #Kubernetes #Rust #CloudNative #Buoyant #CNCF #Microservices #Infrastructure #PlatformEngineering #SoftwareEngineering #SenorsAtScale

💬 What's the most complex networking issue you've debugged in a microservices environment? Share your war stories in the comments!

You can never build anything faster than your slowest database query. In this episode, Tyler Benfield, Staff Software Engineer at Prisma, breaks down everything developers need to know about database performance, from why your queries are slow to how Prisma scales Postgres to handle billions of requests on bare metal infrastructure.

Tyler's path into databases started at Penske Racing, writing trackside software for NASCAR pit stops, and eventually led him deep into query optimization, connection pooling, and building Prisma Postgres from the ground up. We cover the most common ORM anti-patterns, why indexes are the single biggest performance lever most developers ignore, how Prisma Accelerate turns database connections into HTTP calls, and why Tyler thinks the SQL query language itself is fundamentally broken for modern web apps.

Whether you're a frontend developer afraid to touch the database or a backend engineer scaling past your first million users, this conversation is packed with practical, immediately actionable advice.

🔸 Key Topics:

- ORMs vs raw SQL vs query builders and when to use each

- The most common Prisma anti-patterns that tank your app performance

- How database indexes actually work (the address book analogy)

- Connection pooling, serverless runtimes, and the problem Prisma Accelerate solves

- Scaling Postgres on bare metal with memory snapshots and scale-to-zero

- Per-query pricing and why Prisma charges differently than other providers

- NoSQL vs SQL and when Postgres can handle both

- Why SQL is a bad query language for nested relational data

- The future of AI agents and databases, MCP servers, and ephemeral environments

🔗 FOLLOW TYLER

💼 LinkedIn: https://www.linkedin.com/in/tylerbenfield/

🐦 X/Twitter: https://x.com/rtbenfield

🦋 Bluesky: https://bsky.app/profile/rtbenfield.dev

🌐 Website: https://tylerbenfield.me

🎙️ FOLLOW & SUBSCRIBE

📸 Instagram: https://www.instagram.com/senorsatscale/

📸 Instagram: https://www.instagram.com/neciudev

🎙 Podcast URL: https://neciudan.dev/senors-at-scale

📬 Newsletter: https://neciudan.dev/subscribe

💼 LinkedIn: https://www.linkedin.com/in/neciudan

💼 LinkedIn: https://www.linkedin.com/company/señors-scale/

📚 ADDITIONAL RESOURCES

- Prisma ORM: https://www.prisma.io

- Prisma Postgres: https://www.prisma.io/postgres

- The Design of Everyday Things by Don Norman

- The Design of Future Things by Don Norman

- Designing Data-Intensive Applications by Martin Kleppmann

- Aaron Francis (database education): https://aaronfrancis.com

#Prisma #Postgres #DatabasePerformance #ORM #TypeScript #ServerlessDatabase #ConnectionPooling #SQLOptimization #WebDevelopment #SoftwareEngineering #FullStack #DatabaseIndexes #SenorsAtScale

💬 What's the worst database performance issue you've ever debugged? Share your war stories in the comments!