Intelligence Studio: Self-Hosted AI Evaluation, Governance, and Human-in-the-Loop Review A developer released Intelligence Studio v1.1.0, a self-hosted commercial platform for AI evaluation, governance and human-in-the-loop review, announced on Dev.to. The platform combines weighted scoring rubrics, mandatory written justification for sensitive decisions, CSV dataset benchmarking and audit trails, built on Python/FastAPI, React + TypeScript + Vite, PostgreSQL 16 and Redis 7, deployed as eight Docker Compose services on a recommended 4 vCPU, 8 GB RAM, 50 GB SSD spec. It is sold as a self-hosted license with full source code rather than a SaaS subscription, so customer data stays on the buyer's own infrastructure. Intelligence Studio: Self-Hosted AI Evaluation, Governance, and Human-in-the-Loop Review AI systems ship fast. Verifying their output — and keeping a human accountable for sensitive decisions — usually does not. Intelligence Studio is a self-hosted enterprise platform that brings AI evaluation, quality assurance, governance, and audit into one environment. You run it on your own infrast ructure, and your data never leaves it. What it does: Structured AI evaluation — build evaluation cases with deterministic, weighted scoring rubrics, assign analysts and reviewers, and run cases individually or through background workers. Human-in-the-loop governance — sensitive decisions require human review. Approve, reject, or request changes with mandatory written justification, and keep a full trace of every decision. Dataset benchmarking — import CSV records with validation and duplicate detection , ingest through authenticated webhooks, and run repeatable evaluations against labeled datasets. Enterprise security — multi-tenant PostgreSQL 16 with Row-Level Security, RBAC Admin / Analyst / Reviewer / Viewer , TOTP MFA, refresh-token rotation with reuse detection, distributed rate limiting with Redis 7, ClamAV scanning on uploads, and structured JSONL security event logs with SIEM webhooks. Analytics & reporting — executive dashboards, approval-rate monitoring, anomaly detection, trend analysis, and CSV exports. Backup & operations — automated PostgreSQL backups with SHA-256 integrity manifests, retention/cleanup routines, and restore procedures. Stack: Python / FastAPI backend, React + TypeScript + Vite frontend, PostgreSQL 16, Redis 7, Nginx with HTTPS/TLS, deployed via Docker Compose eight services . Recommended production spec: 4 vCPU, 8 GB RAM, 50 GB SSD. This is a commercial self-hosted license v1.1.0 , not a SaaS subscription — you get full source code, deployment scripts, documentation, and security reports, and you own the infrastructure. If your team needs transparent, auditable AI evaluation with humans kept in the decision loop, I'd love your feedback. Questions and criticism welcome in the comments. Key Takeaways - •What it does: Structured AI evaluation — build evaluation cases with deterministic, weighted scoring rubrics, assign analysts and reviewers, and run cases individually or through background workers. - •This story was reported by Dev.to , covering developments in the dev space. - •AI advancements continue to reshape industries — read the full article on Dev.to for complete coverage. 📖 Continue reading the full article: Read Full Article on Dev.to → https://dev.to/capcryo/intelligence-studio-self-hosted-ai-evaluation-governance-and-human-in-the-loop-review-33mk