deployment

Cursor Origin now allows users to host their repositories and deploy to Vercel, with Vercel integration and the service itself hosted on Vercel.

SGLang has updated its deployment recipes for the Qwen3.8-27B model on RTX 5090 and RTX Pro 6000 hardware, adding variants for different configurations with tuning options.

The article discusses the key factors, such as reliability, error handling, and transparency, needed to trust AI agents for real business work, beyond just model intelligence.

A social media post asking practitioners about unexpected challenges and untested issues when deploying AI agents to production, highlighting gaps between development and real-world use.

Santiago Valdarrama highlights an end-to-end system for building, evaluating, deploying, and monitoring specialized AI models, noting that enterprises are willing to pay for custom small models despite the popularity of foundation models.

OpenShip is an open source Platform as a Service (PaaS) that allows deployment on cloud or on-premises infrastructure without vendor lock-in, managing builds, deployments, domains, SSL, monitoring, and more.

This paper evaluates 19 unsupervised anomaly detection models on the BowTie manufacturing dataset, finding performance less stable than benchmarks suggest, and presents a deployed human-in-the-loop framework for manufactured-part inspection.

A robotics practitioner explains that the bottleneck in robotics deployment isn't hardware or models but knowing whether deployments will keep working, focusing on feasibility, reliability, and economics.

LangChain announces the public beta of Managed Deep Agents, a managed runtime for deploying and scaling Deep Agents without managing infrastructure. It supports Python/TypeScript, local testing, one-command deployment, and integrates with LangSmith for production features like persistence, sandboxes, and evals.

This paper studies whether prompt-side agent playbooks (procedures distilled from prior runs) transfer across settings, finding that frozen transfer works only conditionally depending on compatibility with target deployment, with mixed results on ALFWorld, TAU2-Bench, and XBench-DeepSearch.

A technical essay exploring how the conventional web server deployment model (TLS termination, reverse proxying, static file serving, caching) breaks down for hobby-scale apps that must be easily self-hosted, leading to inefficiency and added complexity.

Microsoft Research's Atlas Playbook provides frameworks and tools for designing, deploying, and evaluating human-centered AI systems across diverse cultural contexts, based on fieldwork in Kenya and India.

An analysis of 10,000 voice AI calls reveals key failure points: STT error rates, first-8-second chaos, interruption handling, extended silence, tool call latency, LLM failures, and broken escalation — offering practical insights for agent builders.

An educational guide that breaks down each stage of a production deployment—builds, artifacts, database migrations, health checks, rolling updates, and rollbacks—and discusses when to use a PaaS versus running your own deployment infrastructure.

This repository provides configuration, patches, and tuning to run the DeepSeek V4 Flash 304B checkpoint on a single AMD MI300X in production, achieving 168 tok/s decode without quantization. It includes correctness overlays for vLLM ROCm, AITER tuning tables, and a hybrid KV cache strategy.

This arXiv paper presents a unified LLMOps architecture for real-time, enterprise-ready LLM deployments, integrating data ingestion, continual learning, RAG, and feedback loops. It introduces components like AIPO, STAR+FAR, and SAGE to address knowledge staleness, hallucination, and latency-cost trade-offs in regulated sectors.

A software engineer discusses when feature flags make sense, such as for A/B testing and complex deployments, and cautions against overusing them when teams have control over their deployments.

This paper studies quantization trade-offs for deploying translation models like EuroLLM and Hy-MT2 on a single GPU, showing that combining document chunking with W4A8/W8A8 quantization improves latency-throughput trade-offs while introducing a document-level evaluation to assess long-context dynamics.

Discussion of which platforms truly help enterprises deploy and monitor AI agents at scale, evaluating real-world utility beyond hype.

This talk explains how to use Nix to achieve full reproducibility from development environments, builds, to cloud deployments, and showcases Antithesis's Nix-based "command set" framework as a replacement for messy scripts.