Contact person: Object Recognition & Desktop Fixes

Phone:Show

Email:Show

Location: Pescara, Italy

Budget: Recommended by industry experts

Time to start: As soon as possible

Project description:
"Our video-commerce marketplace is live in staging, built with React on the front end and Python on the back. Before we push to production I need a sharp pair of hands to slot in an object-recognition layer and tidy a few loose ends in the codebase. Namely, the UTM tracking from embedded videos, video performance data sync to user dashboards, affiliate platform to disperse payouts for video performance as specified in the backend.

The recognition service must label objects in real time while a video plays, tag them so they are searchable later, and—most importantly—pipe recognised products into an on-screen product carousel that viewers can click straight through to POS. The CMS already stores product inventory and data; the task is to capture the detection results, map them to existing SKU records, and surface them in the UI without noticeable latency.

Two housekeeping items also block release:

• The “Upload Video” button should take local files, store them in the user’s “My Videos” area (currently nothing lands there), and refresh the list immediately after upload.
• Authentication occasionally rejects valid tokens, so sessions drop at random. A quick audit and patch of that flow is essential, along with fixing the “No LSP Diagnostics found” warning cluttering our dev logs.

I’ll give you access to the private Git repo, env files, and staging server. Deliver a clean PR that:

1. Integrates the object-recognition feature end-to-end.
2. Restores reliable video uploads.
3. Stabilises authentication and removes stray diagnostics.
4. Includes concise setup notes and any new requirements in [login to view URL] / package.json.

Once the updated build passes my smoke tests on staging, we’re good to deploy." (client-provided description)