Issue #167 2 min read

AI Engineering Signal #167

Google DeepMind releases AlphaGenome 2, predicting how DNA variants affect cell-type-specific gene regulation

Share

Signals

Google DeepMind releases AlphaGenome 2, predicting how DNA variants affect cell-type-specific gene regulation

teams building genomic ML pipelines should evaluate this as a new baseline for variant-effect prediction before committing to existing architectures.

Web

Huawei releases MACaQ open-source quantizer for local inference

benchmark against Unsloth GGUFs before locking in your local serving stack.

GitHub

Voltage-glitching converts Nvidia mining card into A100-80G

cheap A100 capacity is now a real option; audit procurement assumptions for local inference builds.

Web

Claude Code "don't ask again" leaking secrets into public repos

audit all Claude Code sessions for exposed credentials before next deploy.

Reddit

Google ships Gemini Decisions API for fast structured outputs

routing decisions and classification tasks now have a dedicated low-latency endpoint to evaluate against current solutions.

Simon Willison

Apple building native Mac app for local model execution

local inference deployment assumptions on Apple Silicon may shift; watch for hardware-accelerated routing changes.

Web

Critical RCE vulnerability found in Apache HTTP Server

any LLM serving stack behind Apache needs emergency patch assessment now.

Web

Get signals like this in your inbox

Daily AI engineering intelligence. No noise.

[ Subscribe ]

The Take

The cheapest inference gains this week came from hardware exploits and open-source quantization, not cloud price cuts — while the most urgent risk is a credential leak in a widely-used coding agent. The infrastructure layer is where the real action is, and it is moving faster than most deployment security postures.

Subscribe

Unsubscribe any time.

Related Signals