Data Science UA is a service company with strong data science and AI expertise. Our journey began in 2016 with the organization of the first Data Science UA conference, setting the foundation for our growth. Over the past 9 years, we have diligently fostered one of the largest Data Science & AI communities in Europe.
About the role:
We are looking for a Computer Vision / Machine Learning Engineer to develop offline CV models for industrial visual inspection.
Your main task will be to design, train, and evaluate models on inspection data in order to:
- Improve discrimination between good vs. not-good samples;
- Provide insights into key defect categories (e.g., terminal electrode irregularities, surface chipping);
- Significantly reduce false-positive (overkill) rates;
- Prepare the solution for future deployment, scaling, and maintenance.
Key Responsibilities:-
Data Analysis & Preparation
~Conduct dataset audits, including class balance checks and sample quality reviews;
~Identify low-frequency defect classes and outliers;
~Design and implement augmentation strategies for rare defects and edge cases.
- Model Development & Evaluation
~Train binary (good vs. not-good) and limited multi-class deep-learning models on inspection images;
~Use modern computer vision / deep learning frameworks (e.g., PyTorch, TensorFlow);
~Evaluate models using confusion matrices, ROC curves, precision–recall curves, F1 scores and other relevant metrics;
~Analyze false positives/false negatives and propose thresholds or model improvements.
- Reporting & Communication
~Prepare clear offline performance reports and model evaluation summaries;
~Explain classifier decisions, limitations, and reliability in simple, non-technical language when needed;
~Provide recommendations for scalable deployment in later phases (e.g., edge / on-prem inference, integration patterns).
Requirements:
Must-have:
- 1-2 years of hands-on experience with computer vision and deep learning (classification, detection, or segmentation);
- Strong proficiency in Python and at least one major DL framework (PyTorch or TensorFlow/Keras);
- Solid understanding of:
- Image preprocessing and augmentation techniques;
- Classification metrics: accuracy, precision, recall, F1, confusion matrix, ROC, PR curves;
- Handling imbalanced datasets and low-frequency classes;
- Experience training and evaluating offline models on real production or near-production datasets;
- Ability to structure and document experiments, compare baselines, and justify design decisions;
- Strong analytical and problem-solving skills; attention to detail in data quality and labelling;
- Good communication skills in English (written and spoken) to interact with internal and client stakeholders.
Nice-to-have:
- Experience with industrial / manufacturing computer vision (AOI, quality inspection, defect detection, etc.);
- Familiarity with ML Ops / deployment concepts (ONNX, TensorRT, Docker, REST APIs, edge devices);
- Experience working with time-critical or high-throughput inspection systems;
- Background in electronics, semiconductors, or similar domains is an advantage;
- Experience preparing client-facing reports and presenting technical results to non-ML audiences.
We offer:
- Free English classes with a native speaker and external courses compensation;
- PE support by professional accountants;
- 40 days of PTO;
- Medical insurance;
- Team-building events, conferences, meetups, and other activities;
- There are many other benefits you’ll find out at the interview.
