Northvolt

Production Machine Vision Engineering

Machine Vision Engineer / Feb 2024 - Jun 2025

Delivered production machine-vision work across line-scan and area-scan imaging, calibration, lighting validation, OpenCV-based analysis, C++/Python debugging, and industrial vendor platforms under real manufacturing constraints.

Line-scan imagingArea-scan imagingCalibrationOpenCVC++ / PythonVendor vision platforms

At a glance

Technical Overview and Key Facts

Role: Machine Vision Engineer
System type: Production machine-vision engineering
Core stack: Line-scan, area-scan, OpenCV, C++, Python, vendor platforms
Constraints: Image quality, calibration behavior, lighting stability, production debugging
Delivered: Inspection analysis workflows, calibration checks, vendor-platform review, debugging support

Problem

Industrial inspection fails quickly when image quality, calibration, lighting, and acquisition behavior are treated as secondary details. The challenge was to support machine-vision workflows that stayed understandable, debuggable, and reliable under real production variation.

My Role

Worked hands-on across industrial imaging setup, calibration, line-scan and area-scan inspection, vendor vision platforms, and C++/Python/OpenCV workflows. Focused on production reliability, image-quality validation, and practical debugging rather than lab-only proof-of-concepts.

Impact

Strengthened the production reliability of machine-vision work by connecting imaging fundamentals, calibration discipline, algorithmic debugging, vendor-platform behavior, and infrastructure awareness into one practical engineering workflow.

System Architecture

Acquisition layer: line-scan and area-scan sensors captured production imagery under real inspection conditions.
Imaging & calibration layer: lighting, exposure, sensor configuration, and calibration stabilized the input before inspection logic was trusted.
Analysis layer: OpenCV, C++, and Python workflows supported inspection validation, measurement checks, image review, and debugging.
Vendor platform layer: Keyence, Cognex, and Dr. Schenk platforms provided production-facing inspection configuration and operator-facing review tools.
Production integration layer: inspection outputs were reviewed and connected into wider production workflows where reliability and traceability mattered.

Tools

OpenCVC++PythonKeyenceCognexDr. Schenk

Infrastructure Touchpoints

Production vision has to fit the wider engineering environment.

Where machine-vision systems touched modern production infrastructure, the work also involved CI/CD, AWS CLI, Kubernetes, and IoT contexts so imaging, debugging, and review workflows could fit the wider engineering environment.

CI/CDAWS CLIKubernetesIoT workflows

Production Constraints

Stable acquisition across line-scan and area-scan inspection conditions.
Lighting and exposure behavior that remained usable under production variation.
Calibration checks before measurement or inspection decisions were trusted.
Vendor-platform configurations that stayed understandable during debugging.
Inspection outputs that engineering and production stakeholders could review.
Practical workflows that worked outside controlled lab conditions.

What I Delivered

Configured industrial imaging setups for production inspection.
Supported both line-scan and area-scan acquisition workflows.
Built calibration and image-quality validation routines.
Developed OpenCV, C++, and Python tools for inspection analysis and debugging.
Worked across Keyence, Cognex, and Dr. Schenk platforms.
Helped separate true algorithm issues from lighting, calibration, acquisition, and configuration problems.

Validation & Commissioning

Commissioning is part of the architecture.

Commissioning was treated as part of the machine-vision workflow, not a final handoff step. Validation focused on image quality, calibration behavior, inspection visibility, vendor-platform configuration, and debugging usefulness under production-like conditions.

Image acquisition behavior.
Line-scan and area-scan visibility.
Lighting and exposure stability.
Calibration behavior.
OpenCV analysis output.
Vendor-platform configuration review.
Production debugging usefulness.

Implementation Highlights

The useful work sits between camera signal and production decision.

Separated image-quality issues from algorithmic issues before changing inspection logic.

Validated calibration and setup before trusting measurement output.

Treated vendor platforms as production tools, not black boxes.

Next Step

Need a vision system that survives production?

Email Umair Open LinkedIn View more systems