Hackathon Portal
AI Tinkerers - Prague
Team

Lattice

Project Concept

No description has been added yet.

Entry

Status: Submitted

Last saved: May 09 at 6:59 PM CEST

Team Roster

You must be registered for the event to view the team message board.

Daniel Zarzecki Team Lead RSVP Approved

Full-Stack / AI developer at Self employed
Solo build. Designed and shipped end-to-end: • Frontend: Vite + React 19 + TypeScript + Tailwind v4 single-page app with two routes • Generative UI runtime: custom sandboxed-iframe renderer with a tiny binding contract (data-field two-way binds, data-bind reads state, data-action button events, data-each loops over data sources injected synchronously into iframe head so agent-emitted inline <script> tags can compute aggregates and render charts on first paint • Agent: Hono server proxying to NVIDIA build.nvidia.com (Llama 3.3 70B via OpenAI SDK, JSON mode, 4096 max tokens, bracket-balancing parser fallback for any truncated output) • Drag-and-drop CSV/JSON/TXT ingestion → DataSource widgets with sortable/filterable fullscreen inspector • Explicit per-widget wiring: drag from a DataSource's orange output port to a Container's blue input port → only wired sources are exposed to that widget's iframe AND included in the agent's prompt context • Per-widget version history: every agent edit and user edit creates a snapshot; click any widget to open detail panel with timeline; revert to any prior state; fork a snapshot into a parallel widget for branching exploration • Drag widgets by header to reposition (camera-aware coord conversion); canvas-only zoom that intercepts browser zoom shortcuts • 6 synthetic CSVs (sales, support tickets, web analytics, products, employees, user events, expenses) generated via seeded RNG for repeatable demos
I have been self employed in AI industry for the past 4 years
Product review, Technical architecture, Knowledge sharing with peers, Community and friendships, Investors, World models for game development, Level design, AI integrations, Reinforcement learning,
World models for game development & level design