#llmops

Point your existing SDK at one URL and reach every LLM vendor — with real failover, not a try/except. One static Rust binary.

fak — the Fused Agent Kernel: one Go binary that turns a tool-using agent (Claude Code, Codex, Cursor, any OpenAI/Anthropic/MCP client) into a managed agent: cache-stable model traffic, context compaction + crash resume, nanosecond tool-call policy, local GGUF serving with SSD expert offload.

Plumbline — a self-learning, customer-value-governed agile AI agent team for Claude Code. 87 subagents + skills, TDD defense-in-depth gates, Kaizen retros, a four-body adversarial council, and an empirically benchmarked QA harness. Does it hang true?

git diff for your agent's context window. See exactly what your LLM saw — turn by turn, block by block.

EvalBot — local-first chatbot security & quality evaluation (FastAPI + Next.js). Evaluate chatbot answers against your own docs & guidelines with ML/NLP + AI-judge scoring. Apache-2.0.

OS-level safeguards layer for AI agent loops — runaway-kill, timeouts & resource limits so coding agents (Claude Code, Cursor, Codex, Antigravity) in YOLO mode can't burn tokens or freeze your Mac

The trust control plane for enterprise AI agents — real-time policy verdicts in under 800ms and litigation-grade evidence of every decision. Pharos decides. Pharos proves.

Case study: unified observability + evaluation for automation and conversational AI agents. Versioned error catalog, multi-signal confidence, cost meters.

GenAI Career Roadmap 2026 🚀 | AI Job Paths & Skills Guide

Enterprise multi-cloud GenAI agent gateway — AWS Bedrock + Azure AI Foundry integrations, policy-based routing, guardrails, cost control and CI eval gate

Open-source enterprise AI agent platform for centrally governing agents, tools, memory, RAG, approvals, and durable workflows with FastAPI, LangGraph, LangChain, and LangSmith.

TokenFuse — runtime control for AI agents: per-run budgets, loop detection, burn forecast, kill-switch. Observability shows the fire; TokenFuse is the automatic extinguisher.

Hermes Agent starter kit for safe scheduled, stateful AI-agent loops

Detect when Claude silently falls back from Fable 5 to Opus 4.8, log it to a local tamper-evident ledger, and keep your work on the frontier model.

A Rust-based PostgreSQL extension that makes relational query optimizers natively aware of LLM token costs, semantic entropy, and latency budgets.

Read-only first-pass incident triage on the Claude Agent SDK + MCP: classifies an alert, retrieves the matching runbook, and proposes a cited first response — or abstains to a human. Measured over four runs: 100% abstention with 0 missed escalations, ~$0.014 per incident.

Local CLI to turn failed AI agent traces (wrong tool, bad args, over-refusals) from LangSmith/Langfuse into clean SFT/DPO fine-tuning data

Vendor-neutral platform for risk-based, evidence-driven and verifiable AI governance, from intake and conditional approvals to runtime assurance.

Runnable Spring AI 2.0 demo — govern every LLM call, MCP tool call and agent-to-agent hop through the self-hosted Dvara control plane. Approval gates, PII blocking, hash-chained audit.

Spring AI starter for Cycles — runtime budget and action authority for Spring AI agents

AI AgenticSafe - stop your AI agent from breaking production at 3 a.m. 8 battle-tested doctrines, 3 real post-mortems, zero dependencies. By Yacine Mahboub, Founder of WEVIA.

Self-hostable LLM inference gateway: policy routing with fallback chains, circuit breakers, semantic caching, per-tenant cost accounting, and Prometheus metrics.

把一台 16G 的 Mac 跑成 Agent 生产系统：四 Agent Harness 全栈解剖——一个金融背景非程序员的 106 天工程记录

Enterprise AI Platform — build, deploy, monitor and govern AI applications at scale (RAG, Agents, MLOps, LLMOps, Governance)

AI-powered Enterprise Knowledge Platform for Intelligent Decision Making, Agentic AI, and Enterprise Intelligence.

Enterprise Agentic AI Platform for autonomous pipeline monitoring, root cause analysis, incident retrieval, data quality validation, and self-healing workflow orchestration using LangGraph, FAISS, Vector Search, and Python.

Enterprise financial intelligence AI agent platform with RAG, SEC EDGAR ingestion, FRED macro analysis, SQL analytics, LangGraph orchestration, security guardrails, evaluation, and observability.

Hands-on Generative AI Engineering repository covering LLMs, Prompt Engineering, RAG, Vector Databases, Fine-Tuning, Multimodal AI, AI Agents, LLMOps, Evaluation, Deployment, and production-ready GenAI applications using OpenAI, Hugging Face, LangChain, LangGraph, and modern AI engineering practices.

Offline regression tests and explainable release gates for RAG systems and AI agents.

A thin, CI-gated evaluation gate for RAG & LLM systems — golden set, band-based pass/warn/fail gates, LLM-judge or heuristic scorers. pip install raggate

SwarmPIM is a production-grade, containerized PIM system engineered with Next.js 14 and FastAPI. It orchestrates a multi-agent AI swarm utilizing Gemini and Grok with dynamic fallback routing. The architecture features an embedded Qdrant vector cache, high-speed Polars batch ingestion, and analytical DuckDB storage optimized via Apache Arrow.

Do your agent pipeline's own checks catch planted faults? Measured on LangGraph, CrewAI and AutoGen: median 16.7%. One prompt-level change takes it to 55.0%. Pre-registered spec, Apache-2.0 harness, every raw trace published.

Local-first DevTools for debugging LangGraph checkpoints and agent memory.

OpenTelemetry observability and privacy-aware failure replay for production AI agents.

Agentic AI ops architecture for VC dealflow — Claude + MCP orchestrating CRM, file storage and Notion

GitHub Action + CLI for replayable CI/PR/release gate decisions - zero-config Self-Test turning checks, scanners, and AI-agent signals into deterministic, tamper-evident PASS/WARN/BLOCK records. Read-only, zero dependencies, Apache-2.0.

Self-hosted, OpenAI-compatible AI gateway that splits your control plane (auth, guardrails, budgets, encrypted key vault, cost & observability) from the translation engine - swap Bifrost / LiteLLM / Portkey / Direct per provider at runtime. Own your keys; contain a compromised engine to one in-flight request.