🏆 DeveloperWeek NY 2026 — Tower Challenge WinnerAIDependencyWatch
Catch the signal before the break — real-time dependency health scanner.
Next.js
React
TypeScriptTHE WORK
Selected Work.
Real projects. Real impact. Each one solved a real problem for real users.
15+
Projects
5+
AI Apps
12+
Technologies
2+
Years
Always
Clean Code
Fast
Shipping
🏆 DeveloperWeek NY 2026 — Tower Challenge WinnerAINext.jsReactTypeScript
Real-time atmospheric crisis platformAI
Production-shaped multi-tenant SaaS.NET
DotNetCore
MERNTuringChat
Real-time chat where humans play games against AI to spot the bot.
Next.jsReact
Node.js
AI
FrontendWeatherly
Mobile-first maps & weather dashboard with a context-aware AI assistant.
Next.jsReactTypeScript
AI
FirebaseDEEP DIVE
From Hackathon Idea To A Winning Product.
signal.repair/dashboard
🏆 Winner — Tower Pipeline Challenge, DeveloperWeek New York 2026
Full-Stack · AI Integration · Data Pipeline
DependencyWatch
Role: Solo Developer
Timeline: Hackathon Build
nextjsreacttypescriptmongodbpython
The Problem
Every modern app stands on hundreds of open-source dependencies — a typical package.json pulls in 800 to 1,500 transitive packages. Developers have no early-warning system for when one goes bad: a fresh CVE, a quietly abandoned maintainer, or a supply-chain hijack. The health signals live in completely different places — CVE feeds, GitHub activity, registries, and live community chatter — and no single tool fuses them into one plain-English verdict.
The Solution
Built an agentic scanner that reads a manifest (or a GitHub repo URL) and produces a 0–100 "repair signal" per dependency by fusing four pillars: Security (OSV.dev CVEs), Maintenance (GitHub activity), Live Chatter (Nimble live web search), and Adoption (registry data). A security-first ceiling forces vulnerable packages into the red — matching npm-audit behavior. A Tower Python pipeline fans out one task per dependency in parallel and caches results in an Apache Iceberg lakehouse, while Claude → Groq → a deterministic engine refines each verdict with graceful degradation.
The Outcome
Won the Tower Pipeline Challenge (Data-to-AI) at the DeveloperWeek New York 2026 Hackathon. Each dependency gets a color-coded score, a recommended action (KEEP · PIN · UPDATE · MIGRATE · FORK), and a citable evidence trail — results streaming in worst-first. Supports six manifest ecosystems and ships a CI/CD GitHub Action that fails any PR introducing a critical dependency.
0
Signals Fused Per Package
0
Manifest Ecosystems
0
Repair Signal Scale
Got a project that deserves the same attention?
Start Your Project