AI Camera & Fingerprint Optimization
Budget / Salary$10–30
TypeFreelance project
LocationRemote
Posted2 hours ago
Our mobile app is stable overall, yet two pain-points are blocking us from scaling smoothly.
Fingerprint log overflow
Right now every fingerprint scan is persisted, so the database keeps growing until the server runs out of memory. I only need the latest record for each user; older entries can be dropped, rotated, or off-loaded in a way that keeps the system auditable but lightweight.
Camera response lag
The heaviest delay shows up while we wait for Jina.ai to process the captured image. A user-facing turnaround of no more than 5 seconds feels acceptable, yet some sessions sit far beyond that. The code is already written and in production, so the task is to profile the hand-off to Jina.ai, trim latency, and return a result inside the target window without degrading accuracy.
Key deliverables
• Refactor the fingerprint module so only the most recent template per user remains in the primary database, with automated pruning or archival handled safely.
• Optimise the Jina.ai image pipeline (batching, concurrency, caching, or other proven techniques) until end-to-end response is ≤ 5 seconds on average under typical load.
• Review the existing code, document the changes, and submit a concise performance report plus any configuration tweaks required for deployment.
I will grant access to the repo, sample data, and current monitoring dashboards so you can benchmark before and after. Completion is confirmed once the server runs 24 hours without memory spikes and the camera module returns results inside the agreed SLA.
Fingerprint log overflow
Right now every fingerprint scan is persisted, so the database keeps growing until the server runs out of memory. I only need the latest record for each user; older entries can be dropped, rotated, or off-loaded in a way that keeps the system auditable but lightweight.
Camera response lag
The heaviest delay shows up while we wait for Jina.ai to process the captured image. A user-facing turnaround of no more than 5 seconds feels acceptable, yet some sessions sit far beyond that. The code is already written and in production, so the task is to profile the hand-off to Jina.ai, trim latency, and return a result inside the target window without degrading accuracy.
Key deliverables
• Refactor the fingerprint module so only the most recent template per user remains in the primary database, with automated pruning or archival handled safely.
• Optimise the Jina.ai image pipeline (batching, concurrency, caching, or other proven techniques) until end-to-end response is ≤ 5 seconds on average under typical load.
• Review the existing code, document the changes, and submit a concise performance report plus any configuration tweaks required for deployment.
I will grant access to the repo, sample data, and current monitoring dashboards so you can benchmark before and after. Completion is confirmed once the server runs 24 hours without memory spikes and the camera module returns results inside the agreed SLA.
Apply on Freelancer →
Project sourced from Freelancer.com. Applications happen directly on the original platform — we never collect your data.