Creating Checkpoints by Gaslighting a Postgres Database
In this episode, Ryan Donovan talks with Brian Clark, Director of Product for LakeBase at Databricks, about the challenges of managing PostgreSQL databases when AI agents are the primary creators and users. Clark explains how AI-driven “fan‑out” strategies spawn many short‑lived database instances, leading to costly token spend and infrastructure overhead, and how LakeBase’s branching architecture—using copy‑on‑write and a custom file system that writes to a virtual WAL—makes creating, cloning, and tearing down databases instant and cheap. He draws parallels between this branching model and Git, emphasizing how it restores developer speed that traditional database provisioning has lost. The discussion highlights the shift from manual scripts to AI automation and the need for platforms that can handle massive, transient database workloads efficiently.
Codex Support, Faster Task Creation, and Flaky Test Visibility
In this Semaphore product update, Pete announces that the SEMAI tool has resolved onboarding blockers, accelerated task creation, and now supports full CLI-based task management, eliminating context switches. The SEMAI plugin has been submitted to the Claude Marketplace, and a...
Building an AI-Native CICD Experience with Sem-Ai
In this Semaphore product update, Pete introduces SEM AI, an AI‑native CI/CD platform that lets developers configure and manage pipelines via natural‑language commands directly from their IDE. A live demo shows how a clean repo can be auto‑bootstrapped with a...
Observability and Human Intuition in an AI World
In this live Stack Overflow episode, Honeycomb CEO Christine Yen and Resolve.ai CEO Spiros Zantos explore how observability must evolve as AI-driven code generation compresses the software development lifecycle. They argue that telemetry remains the same—raw data from applications—but its...
Introducing Semaphore for AI Agents
In this episode, Pete introduces Semaphore for AI Agents, an open‑source CLI that lets AI coding assistants interact directly with Semaphore’s CI/CD platform via natural language. The tool provides commands for pipeline diagnostics, flaky‑test detection, CI insights, and workflow analysis,...
Snap’s Secret to Processing 10 Petabytes a Day: GPU-Accelerated Spark | NVIDIA AI Podcast Ep. 298
Snap’s engineering platform head, Prudvi Vatala, explains how the company slashed data‑processing costs by 76% and reduced core usage by 62% by migrating its 10‑petabyte‑per‑day experimentation pipeline to GPU‑accelerated Spark using NVIDIA Spark RAPIDS on Google Cloud. The move delivered...
TD Cafe #016 – Understanding Drupal Caching with Matt and Nic
In this episode Matt discusses his new self‑published book, *Understanding Drupal: A Complete Guide to Caching Layers*, which demystifies Drupal’s powerful caching system. He explains the core innovations of Drupal caching—cache tags, cache contexts, and cache metadata—that enable granular invalidation...
#547: Parallel Python at Anyscale with Ray
In this episode, hosts Michael Kennedy talks with Ray co‑founders Edward Oaks and Richard Law about the origins and evolution of Ray, the open‑source Python framework that powers large‑scale AI workloads—from its birth in UC Berkeley’s RISE lab for reinforcement...
Lessons From an Upgrade to VCF 9
In this episode, host Pete Fletcher and co‑host John Nicholson interview Michele Collatto, a senior IT leader at a New York‑based financial institution, about their migration from VMware Cloud Foundation (VCF) 5.2 to VCF 9.0 across four global data centers....
#546: Self Hosting Apps for Python People
In this episode, Michael Kennedy talks with Alex Kretzmar, head of DevRel at Tailscale and co‑founder of linuxserver.io, about the state of self‑hosting in 2026. They explore why many users are moving their digital lives behind their own walls, highlight...
Rethinking Git for the Age of Coding Agents with GitHub Cofounder Scott Chacon
In this episode, Scott Chacon—co‑founder of GitHub and CEO of GitButler—discusses how Git’s decades‑old command‑line interface, originally built for human developers, is ill‑suited for today’s AI‑driven coding agents. He explains Git’s Unix‑philosophy roots, its lack of a cohesive UI, and...
Talking Drupal #549 – Catching up with the DDEV Team
In Talking Drupal #549, hosts chat with DDEV maintainers Randy Fay and Stas Huk about the latest updates to the DDEV local development tool and its roadmap. They spotlight the DDEV Drupal Contrib add‑on, which streamlines contrib module development by...
You're Spending Six Figures on AI Models. The Bottleneck Is a 4-Minute CI Pipeline — and Nobody's Fixing the Right...
In this episode, the host explores how the real bottleneck in AI development isn’t the cost of models—often six figures—but a sluggish four‑minute continuous integration (CI) pipeline that slows iteration. They argue that the industry is misallocating resources, focusing on...
#544: Wheel Next + Packaging PEPs
In this episode, Michael Kennedy talks with Jonathan Decker (NVIDIA), Ralph Gommers (QuantSight), and Charlie Marsh (Astral) about WheelNext, a set of PEPs that let Python packages declare hardware requirements so installers like UV can automatically select the optimal binary....
The Messy Truth of Your AI Strategies
In this episode, host Ryan Donovan and guest Hima Raghavan, co‑founder and head of engineering at Kumo.ai, dissect the chaotic realities of deploying AI in profit‑driven enterprises, covering issues like pipeline sprawl, shadow AI, and data governance. Hima explains how...