💥 Issue 92: "I ship code I don't read", Exploiting LLMs vulnerabilities, Becoming a Manager, Big Tech Layoff story, ...

"I ship code I don't read", Exploiting LLMs vulnerabilities, Becoming a Manager, Big Tech Layoff story, ...

Feb 04, 2026

“Should we restructure around AI?” Your team isn't the only one asking. (Sponsored)

At QCon London (March 16-18), 75+ senior practitioners share how they’ve navigated the same decisions. Engineers from LinkedIn, Morgan Stanley, Thoughtworks, UBS, and Cisco on AI’s impact on team structure, architecture patterns for production AI, and what the Staff+ path looks like now.

Three days of practitioner-led sessions, honest post-mortems, and the hallway conversations where someone says “we tried that too, here’s what actually worked.”

Save your team’s spot. Group pricing for 3+.

Save your team’s spot.

Happy Wednesday 👋 and a warm welcome to Tech Talks Weekly #92!

I cross-posted this issue to LinkedIn, Bluesky, and Mastodon and would love your support there as always 🙏

This issue includes over 160 new conference talks published in the past week together with 68 new podcasts. That sounds like a lot, so make sure to check out both 🏆 Featured this week as well as 📈 Most-watched talks to find the ones you don’t want to miss.

Get ready and let’s jump right in!

🏆 Featured this week

Here are our top recommendations this week. Watch now or bookmark!

“Keynote: Can you trust your (large language) model? - Jodie Burchell - NDC AI 2025” from NDC AI 2025
Conference ⸱ +200 views ⸱ Jan 28, 2026 ⸱ 00h 54m 26s
tldw: As LLMs and other “black box” models show up in more products, it gets harder to tell when their outputs are actually correct. The talk covers common ways models can mislead you like sounding confident while being wrong, looking good on a benchmark while failing in production, or passing offline tests but breaking after deployment because the data changed. The main message is to treat model results like any other system output: define what “correct” means for your use case, test with real data and real tasks, track errors over time, and set up checks so you notice when things drift or the model starts acting up.
[Sponsored] Voxxed Days CERN 2026 is coming up on Feb 10, 2026 and the speaker lineup has just been announced. It’s a good time to plan how you’ll use your learning budget (if you have one 😉).
“10 tips to level up your ai-assisted coding - Aleksander Stensby - NDC Manchester 2025” from NDC Manchester 2025
Conference ⸱ +4k views ⸱ Jan 29, 2026 ⸱ 01h 02m 51s
tldw: AI coding assistants like Cursor and Claude Code can make you faster but only if you set them up right. This talk gives 10 concrete tips on prompts, context, debugging, testing, and MCP so you (or rather your assistant) can get more productive.
“186: Becoming a Manager” from Programming Throwdown
Podcast ⸱ Feb 03, 2026 ⸱ 01h 27m 30s
tldl: This episode is about what the job actually is when you stop being an IC and start managing. The main point is that managers are judged on team results and keeping good people around, which means hiring well, setting clear goals and doing the unglamorous work like reviews, comp, coaching, and sometimes downsizing. It also covers why you might want the switch (mentorship, relationships, bigger org impact) and why you might not (less coding, less direct technical control, more open ended problems) as well as how to move back to engineering, if needed.
“Introduction to AI Security - Jim Manico - NDC AI 2025” from NDC AI 2025
Conference ⸱ +200 views ⸱ Jan 28, 2026 ⸱ 01h 00m 41s
tldw: AI features add new security bugs like prompt injection, data leakage, and adversarial inputs. This talk explains how models work and what can we do about it.
“Time is an Illusion - Designing Data-Intensive Applications by Martin Kleppman” from Book Overflow
Podcast ⸱ Feb 02, 2026 ⸱ 01h 22m 00s
tldl: This episode talks through why time is hard to reason about across machines, how clock drift and network delays mess with ordering and correctness, and why you often end up trading off consistency, availability, and latency. It also touches Byzantine failures, what consensus is really buying you, and why batch processing and Unix style pipelines still matter when you are building modern data systems.
“Honest Big Tech Layoff Story After 25 Year Career” from The Peterman Pod
Podcast ⸱ Jan 30, 2026 ⸱ 00h 50m 28s
tldl: A long time big tech engineer talks through getting laid off after a 25 year career and what changed in the job over time. He compares earlier consulting work with big tech, why consultants often get blamed for messy code, and how compensation works when a big chunk is equity that can vanish anytime. He also argues layoffs are not a one off event, shares how age affected his willingness to take risks and move around.
“The creator of Clawd: “I ship code I don’t read” from The Pragmatic Engineer
Podcast ⸱ Jan 28, 2026 ⸱ 01h 54m 05s
tldl: A solo developer behind the ~~Clawdbot~~ ~~Moltbot~~ OpenClaw project explains how he’s shipping at a pace that looks like a small company, mostly by leaning hard on Claude and Codex.