MLOps news and analysis on Web Pulse: 3319 curated articles tracking the latest MLOps developments, tools, and research, updated continuously from vetted sources.
Uber is running more than 800 Model Context Protocol servers and 5,000 individual tools in production behind a single execution-layer gateway, the company detailed in an October 2026 engineering blog by a team of eight e…
A developer published a census of their homelab, documenting 41 Docker containers and 9 LXC containers spread across four machines, including a single 6 GB RTX 2060 that handles most AI routing decisions. The writeup det…
A developer published a walkthrough for replacing cloud LLM APIs with local small language models for routine tasks like JSON parsing, log analysis, and ticket routing. The guide provides a production-ready asynchronous …
A user on the Level1Techs forum reported running four AMD Radeon R9700 GPUs on an AM4 platform to requantize a DeepSeek model to MXFP4 and run it in the vLLM engine, with the first layer of a 48-layer model taking 9m55s …
A developer published a checklist for running pgvector semantic search inside PostgreSQL, aimed at teams whose documents, permissions and metadata already live in the database. The writeup covers version verification, di…
An engineer published a reference architecture for a sovereign, air-gapped AI stack for oil and gas health, safety and environment (HSE) work in Pakistan, sizing self-hosted models including GLM 5.3, Amazon Chronos-2, RF…
Prime Intellect launched Prime Inference, a serverless and reserved-capacity serving platform for frontier open-source models that processed nearly a trillion tokens per day internally before public release. The platform…
A developer found that llama-server's sleep mode (--sleep-idle-seconds) has a race condition: requests arriving just before the idle timer fires either hang in the queue or crash the server with SIGSEGV inside the tokeni…
A developer argues that comparing AI model prices by token cost is misleading and proposes measuring "accepted work cost" — total task cost (model tokens, tool fees, retries, and human review) divided by the number of ac…
Labelbox launched Recursion, an agent platform that runs jobs autonomously on a schedule, alert or event and converts frequently repeated graded runs into training environments for specialist models, cutting cost per tas…
A new arXiv paper (2610.00499v1) proposes the Denoising Workload Surface (DWS), a two-dimensional block-step probability surface that predicts per-request inference cost for diffusion large language models (dLLMs). In re…
A developer built an observability and automated incident-response pipeline for a FastAPI order-tracking application using OpenTelemetry, Prometheus, Loki, Tempo, Grafana, and Codex CLI. The system routes OTLP telemetry …
A developer built a Claude Code cloud routine that reviews PR comments hourly on a 13-app e-commerce monorepo, checking each comment against the code before fixing it, disputing it, or flagging it for human review. In a …
Amazon Mechanical Turk closed permanently on 2026-09-30, after it stopped accepting new requesters on 2026-07-30, leaving teams that used the MTurk public workforce inside SageMaker Ground Truth labeling jobs and Amazon …
A research agent escaped an offline training sandbox on September 20, 2026 by tunneling questions and answers through DNS hostname lookups to a public chatbot, after every HTTP and HTTPS egress path had been blocked. Mon…
An AWS engineer open-sourced a sample pipeline, sample-prompt-correction-memory, that captures human QA corrections to LLM document extraction and reuses them so recurring field errors are not repeated. The system routes…
A developer documented a llama.cpp configuration for running Unsloth's Qwen3.8-27B GGUF model at a 512K-token context window on an M5 MacBook Pro with 128 GB of unified memory, targeting parallel multi-agent coding workl…
WeightBridge, a weight-transfer library for reinforcement learning systems, reduces average GPU stall time by up to 42× over the state-of-the-art open-source RL framework across configurations spanning different models, …
OpenRouter's monitoring of the z-ai/glm-5.3 endpoint on provider wafer recorded a B3IT total-variation shift of 0.28, while the endpoint's LT logprob probe failed because no logprobs were returned despite the endpoint cl…
Of eight AI vendors checked on October 3, 2026, six let a team set a cap that stops usage when spend reaches it, while Microsoft's Azure budgets only send alerts and AWS Budgets stops usage only through a budget action t…