Bachelor's, Master's, or PhD in Computer Science, Electrical Engineering, Robotics, Machine Learning, Mathematics, or a related field.
Hands-on experience with deep learning models for computer vision tasks through research, internships, or substantial projects.
Solid understanding of deep learning fundamentals including model training and optimization.
Familiarity with modern vision foundation models and vision-language models (VLMs).
Proficiency in Python and PyTorch, with experience in Git and Linux-based GPU environments.
Strong communication skills, both written and verbal, with a proactive approach to seeking clarification.
Responsibilities
Train and evaluate models for anomaly detection, zero- and few-shot recognition, and semantic segmentation on industrial image data.
Conduct experiments under senior scientists' guidance, including ablation studies and failure-case analysis, and present findings clearly.
Prepare test datasets and maintain data annotation and evaluation pipelines, ensuring data quality.
Review research literature, share insights with the team, and translate ideas into focused experiments.
Develop clear, reproducible Python and PyTorch code, collaborating with engineers to transition models to production.
Benefits
Paid time off (PTO)
Health, vision, and dental insurance
Life and AD&D insurance
401(k) plan with matching contributions
Additional benefits included in the package
Full Job Description
Aicadium is looking for an early-career Computer Vision / ML Scientist who is motivated to conduct applied research that reaches production. You will join our Applied Research team and work alongside senior scientists on anomaly detection, zero- and few-shot recognition, and semantic segmentation, building on 2D and 3D vision foundation models and vision-language models (VLMs). Your models will be used by enterprise customers, and you will receive structured mentorship and dedicated time for skill development. This is a hybrid role based in San Diego, where our U.S. headquarters is located. The ideal candidate has strong fundamentals, learns quickly, and has a project, thesis, or publication that demonstrates both. Candidates who do not meet every requirement below but are strongly motivated by this work are encouraged to apply.
Responsibilities:
Train and evaluate models, and implement validated methods for anomaly detection, zero- and few-shot recognition, semantic segmentation, and related tasks on real industrial image data.
Run experiments under the guidance of senior scientists, including ablations and failure-case analysis, and summarize findings in clear write-ups and team presentations.
Prepare periodic test datasets and help maintain data annotation and evaluation pipelines, including label review, sampling and dataset splits, and data-quality analysis.
Review relevant research literature, share takeaways with the team, and translate useful ideas into small, focused experiments and prototype improvements.
Contribute clear, tested, reproducible Python and PyTorch code, and work with engineers to transition models to production.
Skills & Requirements:
Bachelor's, Master's, or PhD in Computer Science, Electrical Engineering, Robotics, Machine Learning, Mathematics, or a related field.
Hands-on experience training and evaluating deep learning models for computer vision tasks such as object detection, segmentation, or anomaly detection on non-trivial datasets, through research, an internship, or a substantial project.
Solid deep learning fundamentals: model training, optimization, loss design, evaluation, and generalization.
Familiarity with modern vision foundation models and VLMs.
Proficiency in Python and PyTorch, comfort working with Git and Linux-based GPU environments, and a strong grounding in linear algebra, probability, and statistics.
Clear written and verbal communicator who seeks clarification proactively and incorporates feedback effectively.
Nice to Have:
Publications or workshop papers in computer vision or machine learning.
Exposure to 3D vision, such as point clouds, monocular depth estimation, multi-view geometry, or feed-forward 3D reconstruction.
Experience with zero- or few-shot, self-supervised, or open-vocabulary methods.
Experience with real-world (non-benchmark) datasets or annotation workflows.
Benefits
Aicadium has a great benefits package to come with your salary. Benefits include PTO, health, vision, and dental insurance, life and AD&D insurance, a 401(k) with matching, and more!
