MLOps news and analysis on Web Pulse: 3397 curated articles tracking the latest MLOps developments, tools, and research, updated continuously from vetted sources.
Netflix's AI Platform team built an in-house LLM serving stack, running the full pipeline from model deployment through inference inside its existing production environment. The team selected vLLM as the paved-path engin…
AgentLedger, an open-source Python SDK, provides AI pipelines with structured, queryable audit trails for every consequential action, including decisions, reasoning traces, policy triggers, and human review requirements.…
A developer at the École Supérieure Polytechnique of Dakar (ESP-UCAD) has built AILEF, a universal engineering framework for the AI lifecycle, as part of a Master's thesis in Artificial Intelligence and Big Data. The fra…
Thomas, a machine learning engineer with five years of production deployment experience, released Coloma as an open-source verification and tuning layer for self-hosted large language models. The tool includes a vLLM tun…
Multi-agent coordination is failing in production due to lack of shared context, durable state, and cross-runtime visibility, according to a developer. Industry data shows 57% of organizations already deploy multi-step a…
A developer argues that large AGENTS.md files waste context and hurt maintainability, proposing a modular .harness directory structure as a vendor-neutral alternative. OpenAI has publicly moved away from a large AGENTS.m…
Average GPU utilization in production Kubernetes clusters is just 5%, according to the Cast AI 2026 State of Kubernetes Optimization Report, meaning 95 cents of every GPU dollar goes to idle silicon. Fractional GPUs via …
Hugging Face's default cache directory stores gigabytes of model weights in the home folder, causing storage exhaustion on cloud servers. A developer provides a fix using the HF_HOME environment variable to redirect cach…
Average GPU utilization across tens of thousands of production Kubernetes clusters is just 5%, according to the Cast AI 2026 State of Kubernetes Optimization Report. Default Kubernetes scheduling uses LeastAllocated scor…
Logfire's optimizer automatically fixes prompt errors by analyzing production traces, proposing a single evidence-cited edit without rewriting the entire prompt, and grounding every claim in specific runs that users can …
A developer gave an AI agent read and limited write access to their cloud operations for a week, finding that it reduced mean time to resolution from 45 to 20 minutes on routine incidents but was confidently wrong 3 out …
A developer created a 20-line shell script that automatically resumes rate-limited Claude Code sessions using a PROGRESS.md handoff file and retries. The script waits 5 minutes between up to 20 retries, carrying forward …
A developer has built a calibration and abstention benchmark for K8sGPT-backed Kubernetes SRE workflows, testing whether the system can recognize uncertainty and avoid unsafe actions. The benchmark includes 163 labeled i…
A developer built Otterscope, a lightweight observability tool for LLM agents that runs as a single Go binary with SQLite, avoiding the complexity of hosted or multi-service solutions. The tool ingests OpenTelemetry trac…
A developer building a daily content pipeline found that when Claude hit its weekly limit, the pipeline failed to produce a post for 2026-07-14. The developer redesigned the system to fall back to Grok, adding voice-lint…
A developer argues that AI agents need a semantic layer to produce consistent, governed data access. Without it, agents generate conflicting SQL queries, risk data leaks, and lack auditability. The solution is a governed…
Lians, a three-week-old startup, is building a system of record for AI in regulated workflows to reconstruct what an AI system knew, did, and why at decision time, even after underlying facts change. The company highligh…
A developer describes an operating model for running product, architecture, and backend engineering as a single person with one frontend collaborator and an AI coding agent. The model collapses coordination latency by co…
Glad Labs shipped a series of fixes on 2026-07-16, including renaming a misleading function that caused cloud billing leaks, implementing a fallback ladder for video renders to match planned lengths, and fixing a 'frozen…
A developer at Automations Limited built an AI automation pipeline using Python, FastAPI, n8n, and LLMs to eliminate repetitive business tasks. The architecture triggers workflows, processes data through Python services,…