Agentic AI
Tool calling benchmark: the provider is in the score
Fresh tau2 tool calling benchmark data (Oct 1, 2026): a 0.006 USD task next to a 1.28 USD one. How to read the board …
Blog
Agentic AI, MLOps & Cloud Engineering
Production-grade deep-dives on AI agents, MLOps on AWS, document intelligence - and what cognitive science teaches us about building better AI.
MLOps & Cloud
SageMaker async inference: the scale-to-zero trap
Cut idle SageMaker async inference spend without stranding requests. Build a wake-up policy, durable completion handling, and an honest cost model.
MLOps & Cloud
fMRI data pipeline: bounded-memory BIDS to research API
Build an fMRI data pipeline that tracks discarded scans, bounds voxel reads and gates research API access. See the code and capacity math.
Agentic AI
Agent Tool-Call Budgets: Bound Delegation and Latency
Stop delegated agents from restarting your clock. Build a shared request budget, a hard deadline, and traces that expose the remaining gaps.