Predictive Maintenance Integration Engineer Needed -- 2
Budget: $750 – $1,500 USD
I am building an industrial diagnostic software platform focused on predictive maintenance. The platform must pull machine performance data from temperature, pressure, vibration, and acoustic inputs, stream that information through a DAQ layer, and expose it to an AI engine that will calculate health scores and remaining-life estimates in near real time.
Your assignment is to design and implement the end-to-end integration that makes this possible—from field sensors to the inference API. You will decide the best mix of protocols (e.g., OPC-UA, MQTT, Modbus), structure the data pipeline for sub-second latency, and wire the outputs into the model layer (Python-based, currently using PyTorch but flexible). Robust error handling, time-synchronization, and historical storage (SQL or time-series DB) will all be necessary so the data is reliable enough to train and retrain the models continuously.
Key deliverables
• Hardware/software interface that ingests temperature, pressure, vibration, and acoustic data without loss
• DAQ middleware with buffering, validation, and timestamp alignment
• API or direct bindings that feed the AI module live and batch data
• Deployment guide plus concise code documentation
Acceptance criteria
1. Sensor streams appear in the platform dashboard with <1 s latency.
2. A supplied test script can force a fault scenario; the AI module must receive the event within the same sampling window.
3. All code passes my linting/tests and runs inside a Docker container I provide.
If you have proven experience integrating multi-sensor DAQ systems and shipping predictive maintenance solutions, let’s talk schedule and milestones.
Your assignment is to design and implement the end-to-end integration that makes this possible—from field sensors to the inference API. You will decide the best mix of protocols (e.g., OPC-UA, MQTT, Modbus), structure the data pipeline for sub-second latency, and wire the outputs into the model layer (Python-based, currently using PyTorch but flexible). Robust error handling, time-synchronization, and historical storage (SQL or time-series DB) will all be necessary so the data is reliable enough to train and retrain the models continuously.
Key deliverables
• Hardware/software interface that ingests temperature, pressure, vibration, and acoustic data without loss
• DAQ middleware with buffering, validation, and timestamp alignment
• API or direct bindings that feed the AI module live and batch data
• Deployment guide plus concise code documentation
Acceptance criteria
1. Sensor streams appear in the platform dashboard with <1 s latency.
2. A supplied test script can force a fault scenario; the AI module must receive the event within the same sampling window.
3. All code passes my linting/tests and runs inside a Docker container I provide.
If you have proven experience integrating multi-sensor DAQ systems and shipping predictive maintenance solutions, let’s talk schedule and milestones.