Automated AI Code Review Pipeline

You are a DevOps engineer. Build an automated AI code review pipeline integrated into the pull request flow. Tech Stack: [INSERT - Python/TypeScript service + GitHub/GitLab webhooks] LLM: [INSERT - Claude 3.5 / GPT-4o for reasoning, smaller model for triage] Static Analysis: [INSERT - Semgrep, SonarQube, CodeQL, language linters] Code Context: [INSERT - codebase indexing service / symbol graph] Deliver the following modules in order: 1. Architecture: PR events trigger analyzer fan-out (static tools plus LLM reviewers); an aggregator posts inline review comments. 2. Data Model: review finding (severity, file location, rule id, confidence, suggested fix), repository policies, suppressed findings, review history. 3. APIs: webhook receiver, comment/annotation posting, autofix pull-request creation, feedback reactions (accept/dismiss) used for tuning. 4. Review Capabilities: diff-aware security and bug detection, API-misuse analysis using repo-wide context, test-gap analysis, style/convention checks. 5. Policy Gates: blocking rules per branch, false-positive suppression, code-owner routing, confidence threshold for LLM comments. 6. Testing & Evals: historical-PR benchmark with known defects, precision/recall and developer acceptance rate, prompt regression evals in CI.