Machine Learning Engineer - Computer Vision

Rockwell Automation•Mayfield Heights, OH

17h•Remote

About The Position

Rockwell Automation is building FT Analytics VisionAI, a machine vision and visual inspection into the Portfolio. You will be the senior technical authority for the vision models behind that capability - the detection, classification, and anomaly models that decide whether a part passes or fails on a line running at production rate, on hardware sitting next to a Logix controller. This is a hands-on role. You will set the modeling direction, own the evaluation methodology that says whether a model is good enough to ship, and work directly with manufacturing customers on what their defects actually look like. Accuracy targets here are set by a customer's scrap rate and cycle time, not by a benchmark leaderboard. You will build in an AI-first engineering environment - coding agents running inside a harness we own, evals gating AI-generated changes, GitHub Copilot Enterprise and Claude in the daily loop, and MCP-based tooling that lets agents reach real build, test, and telemetry systems under human-in-the-loop review. You will also help define what eval-driven development means for model code specifically, where the regression suite gates a model release the way a test suite gates a build.

Requirements

Bachelor's Degree or equivalent years of relevant work experience.
Legal authorization to work in the US is required. We will not sponsor individuals for employment visas, now or in the future, for this job opening.

Nice To Haves

Typically requires 8+ years of related experience in a software product development environment.
Bachelor's or advanced degree in Computer Science, Electrical Engineering, Applied Mathematics, or a related technical discipline.
Experience taking machine learning models from research through to production deployment
Experience with deep learning for computer vision - detection, classification, segmentation, or anomaly detection
Experience deploying vision models to edge or embedded targets under fixed latency budgets.
Depth in PyTorch or TensorFlow, and in the tooling around dataset versioning, labeling quality, and experiment tracking.
Experience with model optimization techniques - quantization, pruning, distillation, TensorRT or ONNX Runtime.
Direct customer-facing technical experience: scoping applications, setting acceptance criteria, and defending a model's limits.
Exposure to industrial or manufacturing environments, machine vision hardware, or PLC-based control systems.
Experience building evaluation harnesses that let a team ship model changes with confidence.

Responsibilities

Own the modeling approach for visual inspection - architecture selection, training strategy, and the accuracy, latency, and false-reject targets each deployed model must hit.
Define and maintain the evaluation methodology and regression suite that gates a model release, including how labeled data is versioned, sampled, and audited.
Optimize models for constrained edge targets: quantization, runtime selection, and throughput tuning against real camera and controller timing budgets.
Own drift detection, retraining triggers, and production model monitoring across deployed customer installations.
Partner with customers and application engineers on feasibility, data collection strategy, and acceptance criteria for new inspection applications.
Set the technical bar for the ML discipline through design review, mentoring, and written architecture guidance.
Apply agents and evals to the ML workflow itself - dataset triage, failure-mode summarization, experiment scaffolding - and measure whether they actually improved cycle time.