» curated · synthesized
Skip the noise.
Read the signal.
Curated tech news and synthesis for developers and technology professionals.
» Latest posts
2695 postsUnisoc T606/T616 Backdoors Enable Zero-Click Root on Budget Androids
Chained BootROM, modem RCE, and backdoored OEM apps let attackers silently root Unisoc-based budget Android phones with zero user interaction.
Porting Gemma-4 (2B/4B/12B) to AWS Inferentia2: A Field Report
How Gemma-4's mixed attention heads and KV-sharing broke AWS's vendor stack, and the tracing/sharding fix that made Inferentia2 inference work.
What Actually Crosses the React Server Component Boundary
A deep dive into which prop types survive React 19's Flight serializer across the client boundary, and two subtle traps that break builds despite passing review.
397B MoE Model Runs on a Single RTX PRO 6000 96GB GPU
Krasis runtime runs a 397B-parameter MoE model on one RTX PRO 6000 96GB GPU, hitting 2,354 tok/s prefill via CPU-VRAM expert streaming.
Authenticated RCE Exploits Hit Redis via Stream, TDigest, TopK Bugs
Authenticated RCE chain hits Redis 6.2.22 through 8.8.1 via stream NACK double-free, TDigest and TopK module bugs, bypassing two prior CVE patches.
RefluXFS: XFS reflink race lets Linux users escalate to root (CVE-2026-64600)
Qualys details RefluXFS (CVE-2026-64600), an XFS reflink race enabling local privilege escalation to root on default RHEL, Rocky, and Fedora Server installs.
How Bun rewrote 535K lines from Zig to Rust with 64 AI agents
Bun's creator used 64 parallel Claude agents to rewrite 535,000 lines of Zig into Rust in two days. Key lessons for engineering teams.
Beyond login: encrypting data with passkeys and WebAuthn PRF
WebAuthn's PRF extension turns passkey login into an encryption key source; pknotes shows a master-password-free, end-to-end encrypted architecture.
FreeCAD Ported to the Browser: 2.2M Lines of CAD via WASM
Following LibreCAD and OpenSCAD, FreeCAD has been ported to WebAssembly. In four days, the Fable agent used JSPI, a custom Qt build and a WebGL2 shim to fit a massive CAD kernel into the browser.
Why reformatting prompts beats quantization for CPU LLM inference
Measured on a free ARM CPU box: rewriting prompts as label:value fact sheets cut LLM prefill time 6.5x while quantization tricks hurt accuracy.