The CTO Podcast with Fexingo: Technical Leadership, Architecture, and Engineering Org

In this episode, Lucas and Luna dig into Stripe's 2025 architecture overhaul to handle 100 million API calls per day. They explore the specific bottlenecks—from database contention in PostgreSQL to the migration toward a new ledger service and idempotency keys at scale. Lucas explains how Stripe split its monolithic Payments monolith into domain-oriented services, the trade-off between strong cons...

In episode 103 of The CTO Podcast, Lucas and Luna explore the architectural overhaul behind Spotify's podcast recommendation system. They break down how Spotify moved from a music-first collaborative filtering model to a neural network that understands spoken content, processes 500 million daily listening events, and serves 100 million podcast listeners. The conversation covers the shift from spar...

In this episode, Lucas and Luna dive into Stripe's architectural overhaul of its core payment processing pipeline to handle over 100 million API calls per day. They explore the key decisions: migrating from a monolithic Rails app to a service-oriented architecture, the introduction of a custom idempotency layer, and how Stripe used deterministic simulation testing to ensure correctness at scale. L...

In this episode, Lucas and Luna dive into how Datadog re-architected its core observability platform to handle over 100 petabytes of daily data ingestion. They explore the key engineering decisions behind moving from a monolithic time-series database to a distributed, columnar store optimized for high-cardinality queries. Lucas breaks down the trade-offs between write throughput and query latency,...

In this milestone 100th episode, Lucas and Luna dive into how Figma redesigned its multiplayer architecture to meet enterprise security and compliance requirements. They explore the specific trade-offs Figma made when building a permission system that scales to thousands of collaborators on a single file, how the team moved from a flat permission model to role-based access controls (RBAC) without...

In this episode, we dive into Discord's journey of rebuilding its audio engine to handle 200 million daily voice users across gaming, education, and remote work. Lucas and Luna break down the key architectural decisions: moving from a monolithic audio pipeline to a serverless media proxy mesh, implementing adaptive bitrate codecs that respond to network jitter in real time, and designing a fallbac...

Episode 98 of The CTO Podcast breaks down how Cloudflare's engineering team redesigned its authoritative DNS infrastructure to handle over 100 million queries per second during the largest DDoS attacks. Hosts Lucas and Luna walk through the specific decisions: replacing BIND with a custom Rust-based resolver, adopting a new anycast architecture that spreads load across 330 cities, and the caching...

In this episode of The CTO Podcast, Lucas and Luna dive into the engineering challenges behind GitLab's continuous integration platform. With over 5 million CI jobs running daily in early 2026, GitLab had to redesign its job scheduling, artifact storage, and runner orchestration from scratch. We explore the specific architectural decisions: moving from a monolithic Rails scheduler to a sharded Go-...

Lucas and Luna dive into why MongoDB, after years of relying on the WiredTiger storage engine, decided to build a new storage layer from scratch. They explore the architectural trade-offs between B-trees and LSM trees, how the company handled write amplification and compaction during the migration, and what lessons engineering leaders can take away from a storage engine rebuild that touched every...

In this episode, Lucas and Luna explore how Square's engineering team rebuilt its payments processing stack to handle unreliable internet connections. They dive into the specific design decisions behind Square's offline mode — from local transaction caching and conflict resolution to the eventual consistency model that lets merchants process payments even when connectivity drops. The conversation...

Episode 94 of The CTO Podcast. Lucas and Luna examine the engineering behind the UK's Pensions Dashboard programme, which went live in March 2026. They walk through the core technical decisions: a federated data model where the dashboard doesn't centralise pension records but queries over 50,000 provider endpoints via a standardised API, an identity-matching layer that resolves individuals across...

In this episode, Lucas and Luna dive into how Notion rebuilt its sync engine from the ground up to handle over 100 million blocks per workspace. They explore the shift from a naive JSON-based sync to a conflict-free replicated data type (CRDT) architecture inspired by Figma and Roam Research. Lucas breaks down the key decisions: why they chose a custom CRDT over OT, how they solved cache invalidat...

This episode dives into the real technical challenge behind Figma's multiplayer architecture: not just cursor sync, but handling over 100,000 simultaneous vector objects in a single file. Lucas and Luna unpack how Figma moved from a naive OT approach to a CRDT-based engine, why they chose WebSocket over WebRTC for state sync, and how they manage undo across hundreds of collaborators without breaki...

How Uber rebuilt its dispatch engine to handle 30 million daily rides without a single downtime event. Lucas and Luna break down the core architectural shift from a monolithic matching service to a fully distributed, event-driven system that uses probabilistic matching and decentralized state. They discuss the surprising failure mode during the rollout—a subtle race condition in the geohash partit...

In episode 90 of The CTO Podcast, Lucas and Luna unpack the architectural story behind MongoDB's election protocol redesign. In 2023, MongoDB's replica sets could reliably handle about 10,000 writes per second per primary before election latency became a bottleneck. By 2025, the team at MongoDB had rewritten the consensus layer, cutting re-election time from 12 seconds to under 2 seconds under loa...

In this episode of The CTO Podcast, Lucas and Luna dive into Adobe's multi-year effort to bring Photoshop to the browser. They explore the technical challenges of porting a 35-million-line C++ codebase to WebAssembly, the trade-offs between feature parity and performance, and how the team shipped a pixel-accurate version that runs entirely client-side. Lucas explains why Adobe chose WebGL over Web...

In this episode of The CTO Podcast, Lucas and Luna dive into the technical and organizational challenges behind GitHub's migration of over 100 million repositories to a new Git storage engine. They explore why GitHub needed to move away from the traditional filesystem-based storage, the design of their new custom engine called 'GitV2,' and how they maintained zero downtime during the multi-year ro...

In this episode of The CTO Podcast, Lucas and Luna dive into the technical architecture behind Dropbox's sync engine, which handles file operations for over 700 million users across every major platform. They explore the shift from a two-way sync model to a block-level, content-addressable system, and how Dropbox rebuilt its core to handle conflicts between millions of devices simultaneously. Spec...

Lucas and Luna dig into how Skyscanner, the travel meta-search giant, moved 300 microservices from a synchronous REST model to an event-driven architecture over two years. They cover the concrete trigger (a 2022 outage caused by cascading HTTP failures), the decision to adopt Apache Kafka as the backbone, the strangler fig pattern used to migrate incrementally, and the key metric: a 60 percent red...

Lucas and Luna unpack how Palantir redesigned its Foundry ontology to scale from prototype to production across federal agencies. They focus on the architectural shift from a monolithic object model to a modular, policy-enforced ontology layer that can handle classified, multi-tenant government data. Specific numbers: 5,000 concurrent operators at U.S. Army's Project TITAN, 10x reduction in data i...

Lucas and Luna dive into how Netflix rebuilt its content delivery network to serve over 300 million subscribers across 190 countries. They explore the shift from a third-party CDN to Netflix's own Open Connect CDN, the caching strategy that stores popular titles on ISP servers, and the engineering trade-offs between latency, cost, and reliability. Lucas explains why Netflix pre-positions content d...

In this episode, Lucas and Luna dive into the architectural overhaul Coinbase undertook to scale its blockchain node infrastructure from supporting a handful of cryptocurrencies to over 200 networks. They explore the specific challenges of maintaining consensus across Ethereum, Solana, and Polygon nodes, the migration from monolithic to microservices-based node management, and how the team reduced...

In this episode, Lucas and Luna dive into Shopify's 2025 overhaul of its checkout engine, which reduced average payment time from 18 seconds to under 4 seconds. They explore the architectural choices — from server-side rendering to edge-computed state — that made one-click checkout possible across 2.7 million merchants. Lucas explains how Shopify decoupled payment processing from the storefront, a...

In this episode, Lucas and Luna dive into Stripe's latest architectural overhaul: rebuilding its core accounting and tax-computation engine to handle real-time sales-tax, VAT, and GST across more than 130 jurisdictions. They trace the engineering decisions behind moving from a batch nightly ledger to a streaming event-sourced system capable of recalculating tax lines mid-transaction without blocki...

In this episode of The CTO Podcast, hosts Lucas and Luna dive into Stripe's architectural journey from a simple payment API to a global infrastructure handling payments in over 100 countries and 135 currencies. They explore how Stripe tackled regulatory complexity, latency, and reliability at scale — from its early days of processing a few thousand transactions to now handling billions of dollars...