» Tag
ai-infrastructure
14 postsSecuring AI Agents: From Trust to Containment
As AI agents shift from passive models to autonomous actors, the security perimeter moves inward. A look at OWASP-aligned risks and containment strategies.
The New Physics of Columnar Storage: Parquet, Lance, Vortex, Nimble
How AI workloads and modern hardware are reshaping columnar file formats, from Parquet's renovation to new entrants Lance, Vortex, and Nimble.
Meta Rebuilds BLOB Storage to Cut GPU Stalls in AI Training
Meta redesigned its BLOB-storage architecture with unified metadata and direct-to-client streaming to eliminate GPU stalls in AI training.
CommitBrief — AI code reviews, right in your terminal
A provider-agnostic, local-first CLI that reviews your staged changes, a historic range, or a whole GitHub pull request. Zero telemetry, no server. Free and open source.
commitbrief.comBenchmarking Inference Energy Costs of LLM: A LLaMA Study
Researchers benchmark LLaMA model sizes on V100 and A100 GPUs to analyze the energy and compute costs of LLM inference at scale.
Inside NVIDIA Exemplar Cloud: Closing 8-12% AI Training Performance Gaps
How NVIDIA diagnosed 8-12% AI training performance gaps on H100 and GB200 NVL72 clusters using perf, Nsight Systems, SMMU tuning, and NUMA fixes.
America's biggest grid can now order data centers off the grid
PJM's new emergency authority over data center backup power shows AI infrastructure plans must now account for real grid constraints, not just chips.
Relay: A Self-Hosted LLM Gateway With Eval-Gated Smart Routing
Relay is a self-hosted, single-binary LLM gateway with zero telemetry and an eval harness that gates smart routing by default.
Token Economics: Why Your LLM Bill Outgrows the Pricing Page
Why LLM API bills exceed pricing-page estimates: workload ratios, tokenizer variance, caching, batch discounts, and retry overhead explained for engineers.
Apache Parquet in 2026: The Quiet Format Enters Its Loudest Decade
Thirteen-year-old Apache Parquet faces its busiest development era yet, driven by lakehouse and AI demands: variant types, geospatial support, and footer debates.
AI's next phase: from smarter chat to controlled systems
HalluSquatting attacks, Prime Intellect's $130M raise, and DeepSeek's chip plans show AI advantage shifting from model choice to workflow control.