Hacker
Mahesh Babu Metti Reddy
@maheshmettireddy
Projects
- Top 50
FalkorTrace
Graph HacksOct 2026
Demo submission for the WeMakeDevs × FalkorDB "Graph Hacks" Hackathon (Primarily on company brain - but suits all tracks) 🚀 Live Interactive Demo: https://falkor-trace.vercel.app/ 📦 GitHub Repository: https://github.com/mahesh-mettireddy/FalkorTrace 📖 Blog Post: https://falkortrace.hashnode.dev/why-standard-rag-fails-in-production-debugging-and-how-we-built-a-company-brain-with-falkordb Why do standard Vector RAG agents fail when investigating microservice incidents? Because text retrieval is "0-hop" blind—it only looks at the entrypoint API where an alert fired and misses bugs 3 hops away in downstream dependencies, burning thousands of LLM tokens in retry loops. FalkorTrace puts FalkorDB at the core of automated engineering triage: 1. Multi-Hop Graph Traversal (*0..4): Uses FalkorDB sparse matrices to traverse deep microservice routes, identifying breaking code files and git commits in under 1 millisecond. 2. Sandboxed Code Verification: Executes deterministic math checks against live code artifacts instead of hallucinating answers. 3. Closed-Loop Remediation: Live artifact rollback to clean git SHAs directly updates FalkorDB and proves the fix in real time. 4. Version-Aware Agent Memory: Uses minimum applicability fingerprints (tenant, service, input, artifact SHA, PRD version) to achieve 0ms instant memory hits while automatically invalidating stale findings after rollbacks. 5. Company Brain Knowledge Graph: Connects client tickets, microservices, code modules, git commits, PRD specs, owning teams, and on-call engineers in a single traversable graph. Built with FalkorDB Cloud, FastAPI, Python, and Model Context Protocol (MCP - 12 tools). #falkordb #wemakedevs #graphhacks #ai #graphdatabase #rag #agenticai #softwareengineering