PROFESSIONAL WORK · FARCANA
Farcana AI Search
A real-time AI search interface, built in 48 hours. My work combined frontend delivery, concurrent connection handling, and server-state management.
Product & context
The project was a user-facing AI search interface at Farcana. The recorded scope covers the frontend implementation and a workload of 50 concurrent requests.
My responsibility
I implemented the frontend, used WebSocket pooling for concurrent requests, and integrated TanStack Query for server-state management.
Delivery and reported workload
Latency and timeout figures describe the reported 50-concurrent-request workload. CPU and RAM utilization stayed below 70% in that workload.
My engineering work
Interface and real-time responses
Implemented the React search interface and connected it to the WebSocket response flow.
Connection handling
Used WebSocket pooling to handle concurrent requests. The reported workload reached 50 concurrent requests without observed timeouts.
Frontend server state
TanStack Query was the server-state management library in the search stack, alongside the WebSocket transport.
Results in context
The interface was delivered in 48 hours. The performance numbers are workload observations rather than a production service-level commitment.
NEXT PROJECT