Happy Wednesday 👋 and a warm welcome to Tech Talks Weekly #106!
This week, Tech Talks Weekly has crossed 9,700 subscribers and are edging closer to 10,000! I never imagined Tech Talks Weekly would grow this much and that fast. It’s all thanks to you, your support, and engagement, because most of the readers come through the word of mouth, which is incredible.
Thank you everyone 🙏
This week’s issue features lots of really great talks. From Netflix sharing how they use Java AOT in production, to Stripe explaining how coding agents work inside a huge monorepo as well as distributed databases challenges, HTTP/3 migration at DeliveryHero, memory models comparison across five different programming languages, vector search with OpenSearch, and the real cost of context switching between threads and coroutines.
As always, if you found something useful, hit the ❤️ or leave a comment.
Grab a coffee ☕ and let’s jump in!
Join 9,700+ Software Engineers & Engineering Leads who receive a free weekly email with all the recently published conference talks and podcasts. Stop scrolling through messy YouTube/RSS feeds. Stop FOMO. Easy to unsubscribe. No spam, ever.
Here are our top recommendations this week. Watch now or bookmark!
Conference ⸱ +20k views ⸱ May 21, 2026 ⸱ 01h 03m 34s
tldw: The KotlinConf '26 keynote in Munich opens with a clear theme: Kotlin is becoming a full ecosystem for backend, mobile, web, AI and multiplatform development, not just a language. Highlights include Kotlin 2.4 with stabilized context parameters and explicit backing fields, an 18-month security support policy for the standard library and an alpha release of an official Kotlin extension for VS Code. On the AI side, JetBrains is co-leading the new Agent Client Protocol for IDE-to-agent communication and shipping Junie, its coding agent, with dedicated Android support.
https://youtube.com/watch?v=MmwBJbzWbV0 Conference ⸱ +9k views ⸱ May 23, 2026 ⸱ 00h 49m 35s
tldw: Netflix used Project Leyden to dramatically improve startup time of critical services, but the JDK feature itself was only part of the story. The infrastructure they built around it is what made it work. Ian Brown, who leads Netflix's Java Platform teams, and Martin Chalupa, Senior Software Engineer at Netflix, walk through the autoscaling problem that started the project, a primer on Leyden and ahead-of-time caches, how Netflix integrated the AOT archive into existing deployment pipelines, the lessons learned along the way and the measured results.
https://youtube.com/watch?v=4kEh8hxAP4U Conference ⸱ +8k views ⸱ May 26, 2026 ⸱ 00h 18m 54s
tldw: Same prompt, same agent, same model. Without a context engine: 2.5 hours, 20.9 million tokens and code that would have broken the entire system if it was deployed. With one: 25 minutes and a single nitpick before merge. Brandon Walsenuk from Unblocked argues the problem isn't access but understanding: a context engine reasons across your codebase, Slack history, PR patterns and org structure...
https://youtube.com/watch?v=BiG2ssibKGc Conference ⸱ +3k views ⸱ May 23, 2026 ⸱ 00h 23m 08s
tldw: RL Nabors, a developer experience leader and alumna of the React Core Team and W3C, walks through MCP Apps, which bundle HTML, CSS and JavaScript into a single file rendered in an iframe by the agent, and WebMCP, which lets browser agents call your site's existing functions directly using tool name and description attributes on forms you already have.
https://youtube.com/watch?v=LMbeDEQO6QM Conference ⸱ +3k views ⸱ May 21, 2026 ⸱ 01h 08m 44s
tldw: The Spring I/O 2026 keynote in Barcelona brings together Juergen Hoeller, Rossen Stoyanchev, Oliver Drotbohm, Christian Tzolov and Josh Long for an overview of where the Spring ecosystem is heading. Expect Spring Framework 7 and Spring Boot 4, why the team recommends moving directly to JDK 25, Project Loom with virtual threads and structured concurrency, Spring Modulith and modern application architecture, plus Spring AI 2.0 with MCP support, agentic workflows and Java's role in AI.
https://youtube.com/watch?v=pYDMhBVspIY Conference ⸱ +1k views ⸱ May 22, 2026 ⸱ 00h 40m 25s
tldw: Scaling a database sounds like a great idea, until you start counting the ways it can fail. Add more nodes, throw in replication, and suddenly you're explaining to stakeholders why two customers saw two different truths at the same time. Heather Downing, a Developer Advocate with deep experience across enterprise .NET and mobile, covers the fundamentals like replication, consistency and partitions, along with the tradeoffs nobody mentions when they say "just scale horizontally."
https://youtube.com/watch?v=8zgUMgntNIA Conference ⸱ +100 views ⸱ May 22, 2026 ⸱ 00h 39m 44s
tldw: We've all been told coroutines are "lightweight" and threads are "expensive," but what does that really mean at the OS and hardware level? This talk goes beyond the Kotlin APIs to explore the anatomy of a context switch: the syscall, the kernel trap, the scheduler saving registers and the resulting CPU cache invalidation, then contrasts it with the elegant user-space dance of a suspending coroutine.
https://youtube.com/watch?v=Dv9Mx__pUZA Conference ⸱ +100 views ⸱ May 21, 2026 ⸱ 00h 28m 56s
tldw: REST APIs return data; Cap'n Web returns stubs: references to live objects on the server. That single shift changes everything about how you design APIs. Kenton Varda, lead engineer of Cloudflare Workers and creator of Cap'n Proto, shows why passing functions and objects by reference over the network isn't just a nice DX improvement but a fundamentally different security model for AI agents. You'll see TypeScript-native RPC that eliminates client libraries, promise pipelining that collapses three round trips into one without await, server-side lifecycle management that knows when the client is gone.
https://youtube.com/watch?v=6ExqctHGuVw Conference ⸱ <100 views ⸱ May 24, 2026 ⸱ 00h 32m 12s
tldw: HTTP/3 promises better performance, but rolling it out at scale on Android exposes hidden trade-offs. The DeliveryHero engineering team deployed HTTP/3 across Android apps serving millions of users in 18 European and APAC markets, and used a randomized controlled experiment to measure a 51% reduction in connection handshake latency and up to 17% faster p90 home screen response times, alongside unexpected regressions on some secondary screens. The talk shares the full journey: from identifying last-mile mobile networking as the dominant performance bottleneck, through deployment, to diagnosing why some optimizations never materialized.
https://youtube.com/watch?v=mMWzKHwZEZE Conference ⸱ <100 views ⸱ May 22, 2026 ⸱ 00h 42m 24s
tldw: When the team launched their Android banking app, it had one account, one portfolio and one currency, and the architecture was simple because the assumptions were simple. Then the business grew. This is the story of evolving into a multi-account, multi-portfolio, multi-currency platform without starting from scratch, told as a two-perspective narrative: one speaker covers the business drivers, the other the technical reality on Android (what broke, what scaled and what they'd do differently). You'll leave knowing where single-account assumptions hide in your codebase and why complexity is multiplicative, not additive, when you stack these layers together.
https://youtube.com/watch?v=UNlb1AmxLOk Conference ⸱ <100 views ⸱ May 26, 2026 ⸱ 00h 46m 50s
tldw: Jos Roseboom, founder of EasingYou, takes you on a tour of the common and not-so-common performance pitfalls that turn object-relational mapping into an object-relational mess. From data problems to connection mismanagement, he explores real-world issues using the FunFactStore: a sample app where fun facts are sold but bugs are free.
https://youtube.com/watch?v=RjnXUFyXlno Conference ⸱ <100 views ⸱ May 25, 2026 ⸱ 00h 41m 32s
tldw: Chris Ruiz from Stripe shows what happens when AI stops being an occasional helper and starts living inside a real enterprise-scale monorepo. The talk dives into how Stripe is using LLM coding agents to tackle one of modern engineering's hardest problems: keeping an enormous codebase, full of legacy, internal conventions and historical decisions, under control. Chris walks through agent strategies already in production, from autonomous "minions" that handle simple tasks and produce review-ready pull requests from Slack, to Claude Code-based agents capable of executing multi-step migrations that require planning, research and coordination. He also covers how AI is helping dismantle frontend monoliths by guiding teams through the progressive removal of legacy patterns.
https://youtube.com/watch?v=1yDS6ixahVA Conference ⸱ <100 views ⸱ May 25, 2026 ⸱ 00h 44m 29s
tldw: Horacio Gonzalez, Head of DevRel at Clever Cloud, invites you to rediscover the language from a modern angle and shows how PHP 8+ has become a mature, robust platform fully competitive with anything in today's ecosystem. Expect JIT compilation, stricter typing, attributes and a tooling story built around Composer, the PSR standards, static analysis and dependency injection.
https://youtube.com/watch?v=qBVnMb7cvio Conference ⸱ <100 views ⸱ May 25, 2026 ⸱ 00h 36m 54s
tldw: Memory management is one of those things everyone uses daily but few really understand in depth. Jesús Espino, Principal Engineer at Ona and a veteran of open source projects like Taiga, Penpot and Mattermost, demonstrates how C, Rust, Python, Go and Java manage memory in radically different ways. The talk starts with what actually happens on the stack and the heap, then heads into trickier territory: garbage collection, mark-and-sweep, stop-the-world s and the machinery most of us use without ever seeing what runs underneath.
https://youtube.com/watch?v=1bvwkATB9fM Conference ⸱ <100 views ⸱ May 25, 2026 ⸱ 00h 43m 48s
tldw: In modern ad tech platforms, scale isn't a goal: it's the starting point. Every second, these systems make millions of real-time decisions under brutal latency constraints, where a few milliseconds can directly impact revenue. Diana Cibu, DevLead at Criteo, takes you inside one of engineering's most demanding ecosystems and walks through how systems handling over a trillion HTTP requests per day are designed and operated. You'll follow an ad impression end-to-end, from page load to milliseconds-later decision, and see the distributed systems, queues and caches that make it look invisible.
https://youtube.com/watch?v=W4CXOI8s_wU Conference ⸱ <100 views ⸱ May 25, 2026 ⸱ 00h 40m 28s
tldw: Dotan Horovits, OpenSearch Ambassador at AWS, is talking about moving to modern approach to search: from traditional lexical matching to vector search based on semantic similarity. The session introduces vector database fundamentals, shows how OpenSearch evolved into a platform that understands text, image and audio embeddings and walks through what it enables: semantic search, recommendation engines, visual search and systems that actually start to understand content.
https://youtube.com/watch?v=C_JPL8U21lo Found something useful? Leave a comment. Thank you 🙏
Podcast ⸱ May 26, 2026 ⸱ 00h 46m 54s
tldl: Europe's startup ecosystem is maturing rapidly, with companies like Revolut, Lovable and Legora showing that world-class technology businesses can be built and scaled on the continent. The US still dominates venture-backed software with the largest markets, deepest capital pools and most ambitious exit culture, but a growing number of European startups are challenging that narrative. This Software Engineering Daily episode explores what's actually different about the European scene and what it means for engineers and founders watching from either side of the Atlantic.
Podcast ⸱ May 21, 2026 ⸱ 0 days 00:32:30
tldl: Dr. Barry Feigenbaum, an IBM, Amazon and Dell veteran with decades of Java experience, spent time working with Go on microservices and liked it enough to write the book he wished had existed when he made the switch. In this GOTO Book Club episode with longtime colleague Shon Saliga, IBM Storage Evangelist, he walks through the core contrasts: Go is a compiled language that targets a narrower domain than Java but excels at command-line tools and web servers, with smaller binaries, faster startup and dramatically lower container overhead. Concurrency is the headline difference: goroutines are far lighter than Java threads and the channel-based communication model sidesteps many of the problems that make concurrent Java code hard to reason about.
https://youtube.com/watch?v=vw2k5WVPJKE Here's the complete list of all the talks published since the last issue, grouped by conference, and ordered by number of views for your convenience.
AI Engineer Europe 2026 (18), AgentCon Berlin 2026 (1), Android Makers by droidcon 2026 (17), Code BEAM Lite Stockholm 2026 (3), Codemotion Madrid 2026 (28), Conf42 Golang 2026 (20), Conference Talks 2026 (5), Devoxx Poland 2024 (3), GOTO Copenhagen 2025 (3), JCON EUROPE 2026 (4), JSNation 2026 (1), JavaOne 2026 (1), KotlinConf 2026 (1), KubeCon + CloudNativeCon Europe 2026 (1), NDC Sydney 2026 (13), Observability Day Europe 2026 (1), QCon AI New York 2025 (2), Seattle Rust User Group 2026 (1), Spring I/O 2026 (10), YOW! 2025 (1), performance.now() 2025 (1)
Here's the complete list of all the podcasts published since the last issue.
Find the most watched talks from the past 7d, 30d, 90d, 6m, and 12m.
This issue's talks, sorted by view count and organized by topic.
** “From 46% to 90%: Fine-Tuning Tiny LLMs for On-Device Agents — Cormac Brick, Google”** from
** “Your Coding Agent Should Do AI System Engineering — Ben Burtenshaw, Hugging Face”** from
** “Fast Models Need Slow Developers — Sarah Chieng, Cerebras”** from
→ See 41 more AI Engineering talks on Tech Talks Weekly Plus ** “KEYNOTE: Residuality Theory in Software Architecture, Barry O'Reilly | Code BEAM Lite Stockholm 2026”** from
** “Erlang/OTP Update - Björn Gustavsson | Code BEAM Lite Stockholm 2026”** from
** “KEYNOTE: Ericsson - 150 years of innovation - Mike Begley | Code BEAM Lite Stockholm 2026”** from
** “Distributed Databases: What Could Possibly Go Wrong? - Heather Downing - NDC Sydney 2026”** from
** “Disaggregated Intelligence: Scaling AI Agents with Dynamo and Kubernetes”** from
** “Horacio Gonzalez - Rediscovering PHP: Modern Practices Beyond Legacy -”** from
→ See 2 more Databases talks on Tech Talks Weekly Plus ** “DevOps for Android supercharged with AI”** from
** “Unmask hidden JVM app issues w/Observability superpowers! • Marcin Grzejszczak • Devoxx Poland 2024”** from
** “Beyond Dashboards: Cultivating the Observability Mindset - O. Hope Amaechi-Okorie, JSON Schema”** from
→ See 2 more DevOps talks on Tech Talks Weekly Plus ** “Scalable Design Systems for High-Performance Analytics Platforms | Sonali Priya | Conf42 Golang 2026”** from
** “Modularizing a 10-Year Monolith | Victor Lyuboslavsky | Conf42 Golang 2026”** from
** “From Code Correctness to Behavior Validation | Sana Fatima | Conf42 Golang 2026”** from
→ See 17 more Go talks on Tech Talks Weekly Plus ** “Java AOT in Production at Netflix”** from
** “Spring Native: The Future of Fast and Efficient Spring Applications by Alina Yurenko @ Spring I/O”** from
** “Spring I/O 2026 Keynote”** from
→ See 10 more Java talks on Tech Talks Weekly Plus ** “Lobster Trap: OpenClaw in Containers from Local to K8s and Back — Sally Ann O'Malley, Red Hat”** from
** “Scaling Agents on Kubernetes with acpx and ACP — Onur Solmaz, OpenClaw”** from
** “WebAssembly on Kubernetes • Nicolas Frankel • GOTO 2025”** from
→ See 2 more Kubernetes talks on Tech Talks Weekly Plus ** “From Chaos to Cohesion: Building Modular Monoliths in the Real World by Patrick Baumgartner SIO26”** from
** “The Persistence Heavyweight Championship: JPA vs. jOOQ by Pasha Finkelshteyn / Catherine Edelveis”** from
** “Continuous Delivery in a World of Constant Change • Abby Bangser & Dave Farley • GOTO 2025”** from
→ See 24 more Misc talks on Tech Talks Weekly Plus ** “Keynote | KotlinConf ’26”** from
** “Build The Million-Dollar Animation”** from
** “Coroutines vs. Threads: The True Cost of a Context Switch”** from
→ See 16 more Mobile talks on Tech Talks Weekly Plus ** “9 Ways to do Inheritance in Rust — by Carl Kadie — Seattle Rust User Group, May 2026”** from
** “Speed, Safety, Sustainability. Pick 3 with Rust & Serverless - James Eastham”** from
** “Rustam Mehmandarov (Miles) - API = Some REST and HTTP, right? RIGHT?!”** from
** “Skill Degradation: An Empirical Analysis of 400+ AI‑Generated Security Fixes - Pedram Hayati”** from
** “The Rise of Automated Fraud: Architecting Security Against "AI Slop"”** from
** “The Good, The Bad and The Ugly of AI in Security • Olympiu Pop, Steve Poole • Devoxx Poland 2024”** from
→ See 2 more Security talks on Tech Talks Weekly Plus ** “The Reckless Speed of AI Agents: Architectural Amnesia EXPOSED”** from
** “KEYNOTE: Residuality Theory in Software Architecture, Barry O'Reilly | Code BEAM Lite Stockholm 2026”** from
** “The Rise of Automated Fraud: Architecting Security Against "AI Slop"”** from
→ See 3 more Software Architecture talks on Tech Talks Weekly Plus Never miss great conference talks and podcasts again for ** $7 $5/month**.
📁 Search, filter, sort the full archive of talks & podcasts since 2020.
🗄️ Complete category view of new talks & podcasts.
📈 Most-watched talks from the past 7d, 30d, 90d, 6m, and 12m.
Subscribe to Tech Talks Weekly Plus → This issue is free for everyone, so feel free to share or forward it!
Enjoy the weekend ☀️ and see you next week!