MLOps news and analysis on Web Pulse: 3506 curated articles tracking the latest MLOps developments, tools, and research, updated continuously from vetted sources.
Databricks launched Genie ZeroOps at its Data + AI Summit to automate monitoring, investigation, and remediation of data and AI workload issues. The tool, currently in private preview, uses an AI agent to identify anomal…
Nexus Labs enabled speculative decoding in vLLM for a fine-tuned 8B model, achieving a 1.9x throughput gain, but discovered that greedy decoding with a draft model is not bit-identical to greedy decoding without one due …
A developer built an autonomous incident investigation agent called FRIDAY that reduced mean time to resolution by 65% on a platform serving 30+ million end users. The agent, triggered by PagerDuty alerts, first checks G…
A developer outlines the need for observability—metrics, logs, and traces—in autonomous AI systems, providing Python code examples using Prometheus and structured logging to monitor non-deterministic behavior and detect …
A new five-question audit framework helps teams evaluate AI agent harnesses before updating models, preventing failures from mismatched prompts, tool calls, and downstream integrations. The framework focuses on input sou…
ToolSchema Kit, a new open-source repository, provides a reproducible MCP contract drift lab for agent platforms like Cursor and Claude Desktop, enabling teams to detect silent breaking changes in tool schemas. The kit i…
Port has identified seven categories of hidden infrastructure debt that arise when AI agents move from local demos to production in enterprise systems, including integrations, observability, governance, human-in-the-loop…
Sergio Colque Ponce, a software engineering student at Universidad Privada de Tacna, proposes Spec-Driven Development (SDD) as a methodology to improve AI coding agent accuracy by writing specifications before code. He d…
By 2026, AI workloads will consume roughly 40% of enterprise Kubernetes clusters, but the default scheduler is ill-suited for GPU-intensive tasks, leading to 30-45% GPU utilization rates and wasted costs. Platform teams …
TextStack, a .NET-based AI product, has implemented a two-part evaluation system: an offline gate that runs golden dataset evals as opt-in dotnet tests to catch regressions before shipping, and production monitoring that…
AWS has added release management features to its DevOps Agent, now in preview, that automatically assess code changes against organizational standards, identify release risks, and generate tests. The new capabilities aim…
An engineer on Oracle Cloud Infrastructure (OCI) built a scale-to-zero AI inference system on Oracle Kubernetes Engine (OKE) using KEDA to eliminate costs from idle GPUs. The system scales GPU pods down to zero when ther…
Claude Code and Codex, two coding agents, autonomously improved AI agent applications by performing engineering practices such as clustering failure patterns and running ad-hoc evaluations, without relying on specialized…
Oxlo.ai outlines best practices for optimizing large language model performance in production, emphasizing prompt design, model selection, and request architecture. Techniques include deduplicating static content, rerank…
A team of CV engineers deploying a pedestrian detector on a Cortex-A53 board enabled Winograd convolution for a speedup but lost 2.1 mAP due to fp16 overflow in the F(4,3) transform, with small objects dropping 7 points.…
A developer reports that AI-generated code, while fast to produce, creates 'Ghost Implementations'—code that works but lacks justified logic, making maintenance and evolution difficult. After an AI built a UI in two hour…
A developer has been refining KMDS, a framework for building repeatable and auditable machine learning systems, with recent updates focusing on analytical governance. New features include a Feature Advisor service that r…
QuantaMind built a headless CLI tool to shift AI evaluation left into CI/CD pipelines, allowing teams to gate model merges based on agentic reliability thresholds. The tool integrates custom eval JSON collections into th…
A developer argues that AI coding tools should be governed with the same rigor as infrastructure, drawing parallels to DevOps practices like blast radius analysis, auditability, and measurement. The author notes that whi…
A developer from MartinLoop argues that unbounded retry loops in agent systems are the primary driver of cost and trust erosion, not initial failures. The team implemented a control-layer policy with budget caps, max att…