
◈ RepoNavigator AIVisual Developer Onboarding & Interactive Architecture Cockpit"Don't just chat with your repository — visually understand it."🚨 The ProblemWhen software engineers join a new company or contribute to complex repositories, up to 70% of initial onboarding time is spent deciphering folder hierarchies and tracing cross-service communication rather than writing business logic.Traditional documentation suffers from critical flaws:Outdated Text READMEs: Architecture docs degrade over time.Generic AI Chatbots: Standard chat extensions lack spatial topology awareness.High Cognitive Load: New developers spend days grepping files to answer questions like "Where does authentication happen?".💡 Our SolutionRepoNavigator AI transforms any codebase into an interactive visual cockpit. Powered by IBM Bob 2.0 agent intelligence and AST static analysis, it gives engineers instant spatial comprehension:🗺️ Interactive Architecture Map: Dynamic React Flow graph parsing microservices, gateways, and DB layers.🔎 Subsystem Inspector: Slide-out inspector detailing module responsibilities, endpoints, and underlying source files.💬 Topology-Grounded AI Copilot: IBM Bob-powered assistant tracing multi-service lifecycles with clickable @Module badges that navigate the visual canvas.🎓 Personalized Onboarding: Role-tailored onboarding curricula with task tracking and one-click terminal setup commands.📁 Live Ingestion: Ingest any public GitHub repository or multi-service architecture in real-time.⚙️ Built With & IBM Bob 2.0 IntegrationRepoNavigator AI uses a modern high-performance stack: Next.js 16, React 19, Tailwind CSS v4, @xyflow/react (React Flow v12), and Tree-sitter static AST analysis.How IBM Bob 2.0 Powers the Experience:Autonomous Architecture Synthesis: Bob analyzes directory trees and route definitions to generate node topologies and edge protocols.Cross-Service Trace Synthesis: Bob constructs grounded multi-hop execution lifecycles for complex trace queries.Onboarding Generation: Bob synthesizes environment setup, migrations, and architectural checkpoints tailored to the user's skill level. 🚧 Challenges Dynamic Dependency Mapping: Translating unstructured file trees into clean directed acyclic graphs (DAGs) required custom layout algorithms. Contextual Grounding: Preventing hallucinations by grounding Bob with AST symbol extraction and exact file paths. Performance: Achieving 60 FPS graph rendering across a synchronized 3-column UI. 🏆 Accomplishments True Multi-Service Support: Zero-latency architectures including Google Cloud's 10-microservice demo. Interactive Module Badges: Clickable @Module mentions that zoom to nodes on the live canvas. Actionable Onboarding: Cutting Day-1 setup time from days to under 30 minutes. 💡 What We Learned How to effectively leverage IBM Bob 2.0 for architectural reasoning and code graph generation. The power of visual mental models over pure text chat for developer productivity. 🚀 What's Next Live IDE Extensions: Bringing the visual cockpit to VS Code/Cursor. PR Architecture Diffing: Visualizing architectural boundary changes on Pull Requests. Live Distributed Tracing: Connecting OpenTelemetry spans for real-time traffic visualization.
27 Sep 2026

AI Traveler_Planner is a personalized AI-powered web app that turns complex travel planning into a smart, enjoyable experience. Built with Groq’s LLaMA model, it generates detailed itineraries, budget breakdowns, and travel tips based on a traveler’s preferences, interests, and real-time data. What makes it unique is its graph-based profile system using NetworkX, which continuously learns from user behavior to deliver better recommendations. With live flight and hotel integration via Tavily and Booking.com APIs, and a secure email authentication system, it feels like having a personal travel agent that truly understands you. Whether you're a solo backpacker or a family traveler, AI Traveler_Planner adapts your journey to your needs — intelligently and beautifully.
8 Jul 2025