Learning Project / Tools
Findly Now is a from-scratch learning project built with Elixir to explore the language ecosystem and modern AI-assisted development workflows. The development leveraged:
- Claude Code for AI-powered code generation and architecture decisions
- v0 for rapid UI prototyping and component design
- Vibecode for enhanced developer experience and tooling
Recently published on GitHub as an open-source project. Currently in development phase, not yet deployed to production.
The "Why" (Elixir/Phoenix)
I chose Elixir/Phoenix both to leverage the BEAM VM's strengths in concurrency and fault-tolerance and to push myself out of my usual Node.js comfort zone. The idea is to have something that can naturally handle lots of long-lived connections and real-time updates without extra complexity.
Functionality
Findly Now is a domain-specific real-time search and alerting app. Think of it as a way to search and subscribe to fresh opportunities (e.g. remote/tech jobs and freelance gigs) aggregated from multiple sources. Users can filter by stack, location, or tags and get live updates via websockets when new matches appear.
Performance / Concurrency
In internal tests Phoenix handles thousands of concurrent websocket connections easily, and search queries run through a pre-indexed store so they return in tens of milliseconds for typical filters. The goal is to keep the architecture simple—one Phoenix app with channels, PubSub, and a search layer—while still proving out the concurrency and real-time capabilities of Elixir.