<h2>Must know</h2>
<ul>
<li><strong>Embedding model deprecation causes vendor lock-in</strong>: Google deprecated its <code>text-embedding-004</code> model, rendering thousands of stored vectors unusable for a production RAG-based movie recommender. This highlights a critical vendor lock-in issue: different embedding models produce incompatible vector spaces, making migration difficult. The solution involved re-embedding the entire dataset and a blue/green swap, emphasizing the need to design for change when persisting embeddings at scale. <a href="https://app.daily.dev/posts/xhDI1zj4i" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Databricks launches lakebase for AI workloads</strong>: Databricks launched Lakebase, a new PostgreSQL-compatible database designed for AI workloads, aiming to bridge operational database capabilities with their data lakehouse architecture. This serverless OLTP database, built with technology from Neon and Mooncake Labs, supports up to 8TB per instance and includes Postgres 17 with pgvector support, offering features like instant data branching and unified access controls. <a href="https://app.daily.dev/posts/56RHAShlB" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Uber and OpenAI unveil traffic control platforms</strong>: Uber and OpenAI revealed their infrastructure-level traffic control platforms: Uber’s “Global Rate Limiter” handles 80 million requests per second using “probabilistic shedding,” while OpenAI’s “Access Engine” employs a “credit waterfall” mechanism. These systems are designed to manage high-volume traffic and ensure fair access across their respective platforms, showcasing advanced approaches to distributed system resilience. <a href="https://app.daily.dev/posts/OVoWP2oNj" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Good to know</h2>
<ul>
<li><strong>Kubernetes strengthens AI/ML capabilities</strong>: Kubernetes v1.35, codenamed ‘Timbernetes,’ introduces features like workload-aware scheduling and stable in-place Pod resource resize, solidifying its role as an operating system for AI/ML workloads. <a href="https://app.daily.dev/posts/mAyDHBOpw" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Shuru offers local microVM sandbox for AI agents</strong>: Shuru is a new local-first microVM sandbox for macOS, designed for AI agent workloads, providing secure, ephemeral, and reproducible environments without requiring Docker. <a href="https://app.daily.dev/posts/TqgsBjEoV" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>ChatDetector uses LLMs to find RM-API misuse</strong>: ChatDetector, an LLM-powered system, detects resource management API misuse in software by employing a ReAct-inspired framework and two-dimensional prompting, identifying 115 security bugs across six popular libraries. <a href="https://app.daily.dev/posts/UGo4Ea1t8" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Linux kernel 7.0 rc1 released with Rust support</strong>: The Linux kernel 7.0 rc1 release includes cemented Rust support, faster cache clearing, non-disruptive kernel updates, and performance improvements for RISC-V and LoongArch. <a href="https://app.daily.dev/posts/cYP8biu57" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Non-coding tasks hinder developer productivity</strong>: A developer’s account highlights that a 10-second code fix can take over 10 hours due to non-coding tasks like fighting sandboxed environments, <code>CUDA</code> conflicts, and build system issues, which <code>LLM tools</code> currently don’t address. <a href="https://app.daily.dev/posts/OBIqfWI7x" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Rust RwLock underperforms Mutex in read-heavy workloads</strong>: A hands-on investigation revealed that <code>RwLock</code> performed ~5× worse than <code>Mutex</code> for a read-heavy LRU tensor cache in Rust due to cache-line ping-pong, suggesting <code>Mutex</code> might be better for tiny critical sections. <a href="https://app.daily.dev/posts/rFMs5918e" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Microsoft project silica advances glass storage</strong>: Microsoft’s Project Silica achieved a breakthrough in glass-based data storage, extending it to common borosilicate glass with a new phase voxel method and parallel multi-beam writing, suggesting data longevity of at least 10,000 years. <a href="https://app.daily.dev/posts/KqLBpGspa" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
<h2>Bullshit police</h2>
<ul>
<li><strong>Anthropic’s Claude compiler video: more hype than reality?</strong>: Anthropic released a marketing video showcasing Claude autonomously building a C compiler capable of compiling Linux, SQLite, Lua, and Doom. Critics, however, are calling out the “dishonest framing,” suggesting the actual goal was merely a stress test of Claude’s ability to run autonomously for an extended period, not a straightforward demonstration of compiler-building prowess. Because, you know, a marketing video would never exaggerate an AI’s capabilities. <a href="https://app.daily.dev/posts/vGYk0DbTO" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>Neiry’s bio-drones: pigeons with neural interfaces for surveillance</strong>: Neiry, a deeptech company, announced early real-world testing of ‘bio-drones’—live pigeons fitted with neural interfaces, solar panels, and cameras to guide them along preset routes. They claim these birds can cover 400 km daily for infrastructure inspection and search-and-rescue, with “privacy compliance” through on-device processing. Because nothing says “privacy-first” like a pigeon with a brain implant and a camera, flying over your house. <a href="https://app.daily.dev/posts/m7v8MA0Dh" target="_blank" rel="noopener nofollow">Read more</a></li>
<li><strong>MySQL community begs Oracle for a foundation</strong>: The MySQL community is begging Oracle to establish a vendor-neutral foundation, citing declining popularity, missing modern features like native vector search for AI, and a closed-door development process. Oracle’s response? A blog post about improving community engagement. Because a few roadmap updates are definitely going to fix decades of neglect and the complete absence of AI features that everyone else has. Good luck with that. <a href="https://app.daily.dev/posts/UyLIHBdUw" target="_blank" rel="noopener nofollow">Read more</a></li>
</ul>
