Industrial Vision Engineer — Real-Time Edge ML

Conduit; San Francisco (CA); On-site; USD 150,000 - 230,000; Full time

Benefits offered by this job

Equity

Job summary

Conduit is hiring a customer-first CV/ML engineer to drive perception, data science and ML in live factory data streams. You will own the streaming vision system, capturing, classifying and acting on signals in real time on factory hardware.

You will develop end-to-end pipelines from data collection to deployment, train models in PyTorch, and deploy on edge devices you select, collaborating with plant managers for real-world adoption.

Qualifications

• 5+ years of applied computer vision and perception work, including production deployments.
• Strong experience building end-to-end training to inference pipelines and labeling workflows (manual and automated).
• Proficient in Python and PyTorch; familiar with CV techniques and edge deployment.

Responsibilities

• Own customer CV/ML deployments end to end: scope, edge compute, data pipeline, training, deployment, floor validation.
• Build end-to-end pipelines from training to inference: data collection, training, evaluation, deployment, monitoring.
• Stream and process video at industrial volume with multi-camera setups and real-time edge inference.
• Apply CV to industrial problems: defect detection, ergonomics analysis, machine state, material tracking.
• Spec and order hardware to run it; GPUs, edge boxes, cameras, lenses, lighting (can be a $40K compute order).
• Translate customer needs into clear engineering briefs for the SF team and build relationships with plant managers.

Skills

Applied CV

Python

PyTorch

Edge inference

Hardware awareness

Customer-facing

Tools

TensorRT

ONNX

Triton

GStreamer

CUDA

Job description

Conduit is hiring a customer-first CV/ML engineer to drive perception, data science and ML in live factory data streams. You will own the streaming vision system, capturing, classifying and acting on signals in real time on factory hardware.

You will develop end-to-end pipelines from data collection to deployment, train models in PyTorch, and deploy on edge devices you select, collaborating with plant managers for real-world adoption.