Workers AI for Surveillance Guardian
Budget: £250 – £750 GBP
We are looking for an experienced backend / AI engineer (or team) to build the first production version of Workers AI, the intelligence engine behind our commercial CCTV health monitoring product: Surveillance Guardian.
Surveillance Guardian receives snapshots and telemetry from on-site devices. Workers AI will analyse images, process hardware telemetry, compute health scores, and return actionable alerts to our platform.
This project includes:
⸻
Project Scope
1. API Development (FastAPI / Node / Similar)
• Create a secure Workers AI API.
• Endpoints to receive:
• Camera snapshots (already stored by our API server — Workers AI retrieves them via URL).
• Telemetry data from on-site SG Edge devices.
• Endpoints for our platform to fetch:
• Latest camera health status.
• Time-series metrics.
• Alerts.
⸻
2. Ingest & Processing Pipelines
Implement two asynchronous pipelines:
A. AI Image Analysis Pipeline
• Fetch snapshot from our API server (URL-based).
• Preprocess image.
• Run AI/ML models or deterministic algorithms to detect:
• Blur / sharpness
• Noise (banding, lines)
• Occlusion / blocking
• Exposure problems
• Fog/condensation
• Glare
• IR reflection
• Colour cast
• FOV tampering (baseline + SSIM)
• Produce image metrics + issue flags and store them.
B. Telemetry Processing Pipeline
• Normalise and score hardware data:
• Temperature
• Bitrate
• Frame rate
• Network errors
• Motor faults
• SD card health
• Firmware version
• Uptime
⸻
3. Health Scoring Engine
Combine latest image metrics + telemetry to calculate:
• Image Health (0–100)
• Hardware Health (0–100)
• Network Health (0–100)
• Overall Health Score
Return:
• Status (OK / Warning / Critical)
• Primary issue cause
⸻
4. Alerts Engine
Trigger alerts based on rules such as:
• Overheat
• Dirty lens
• Low bitrate
• FOV tamper
• Sudden health drop
• Telemetry out of spec
Alerts must be written into a DB and returned via API.
⸻
5. Storage Layer
Developer may choose appropriate stack (we prefer Postgres + Timescale or Postgres + InfluxDB):
• Timeseries data: metrics & scores
• Metadata: cameras, sites, alerts
• Snapshot references: URLs (no binary storage required)
⸻
Deliverables
• Fully functioning Workers AI backend
• API documentation (Swagger/OpenAPI)
• Dockerised services
• Database schema
• Health scoring logic
• Alerting rules engine
• Deployment guide
⸻
Bonus Points (Not required but helpful)
• Experience working with CCTV, ISAPI, or image processing
• Ability to optimise AI models (ONNX Runtime)
• Predictive analytics (simple regression or GBT models)
⸻
Requirements
• Strong backend experience (Python/Node)
• Experience with queues (RabbitMQ / Redis Streams / Kafka)
• AI image processing knowledge (OpenCV, ONNX)
• Solid database architecture
• Good communication and documentation skills
⸻
About Us
We build intelligent monitoring systems for the fire & security industry. Workers AI will power the next generation of fully automated CCTV system health checks.
Long-term collaboration available for the right developer/team.
Surveillance Guardian receives snapshots and telemetry from on-site devices. Workers AI will analyse images, process hardware telemetry, compute health scores, and return actionable alerts to our platform.
This project includes:
⸻
Project Scope
1. API Development (FastAPI / Node / Similar)
• Create a secure Workers AI API.
• Endpoints to receive:
• Camera snapshots (already stored by our API server — Workers AI retrieves them via URL).
• Telemetry data from on-site SG Edge devices.
• Endpoints for our platform to fetch:
• Latest camera health status.
• Time-series metrics.
• Alerts.
⸻
2. Ingest & Processing Pipelines
Implement two asynchronous pipelines:
A. AI Image Analysis Pipeline
• Fetch snapshot from our API server (URL-based).
• Preprocess image.
• Run AI/ML models or deterministic algorithms to detect:
• Blur / sharpness
• Noise (banding, lines)
• Occlusion / blocking
• Exposure problems
• Fog/condensation
• Glare
• IR reflection
• Colour cast
• FOV tampering (baseline + SSIM)
• Produce image metrics + issue flags and store them.
B. Telemetry Processing Pipeline
• Normalise and score hardware data:
• Temperature
• Bitrate
• Frame rate
• Network errors
• Motor faults
• SD card health
• Firmware version
• Uptime
⸻
3. Health Scoring Engine
Combine latest image metrics + telemetry to calculate:
• Image Health (0–100)
• Hardware Health (0–100)
• Network Health (0–100)
• Overall Health Score
Return:
• Status (OK / Warning / Critical)
• Primary issue cause
⸻
4. Alerts Engine
Trigger alerts based on rules such as:
• Overheat
• Dirty lens
• Low bitrate
• FOV tamper
• Sudden health drop
• Telemetry out of spec
Alerts must be written into a DB and returned via API.
⸻
5. Storage Layer
Developer may choose appropriate stack (we prefer Postgres + Timescale or Postgres + InfluxDB):
• Timeseries data: metrics & scores
• Metadata: cameras, sites, alerts
• Snapshot references: URLs (no binary storage required)
⸻
Deliverables
• Fully functioning Workers AI backend
• API documentation (Swagger/OpenAPI)
• Dockerised services
• Database schema
• Health scoring logic
• Alerting rules engine
• Deployment guide
⸻
Bonus Points (Not required but helpful)
• Experience working with CCTV, ISAPI, or image processing
• Ability to optimise AI models (ONNX Runtime)
• Predictive analytics (simple regression or GBT models)
⸻
Requirements
• Strong backend experience (Python/Node)
• Experience with queues (RabbitMQ / Redis Streams / Kafka)
• AI image processing knowledge (OpenCV, ONNX)
• Solid database architecture
• Good communication and documentation skills
⸻
About Us
We build intelligent monitoring systems for the fire & security industry. Workers AI will power the next generation of fully automated CCTV system health checks.
Long-term collaboration available for the right developer/team.