Job Description
We are looking for a Junior to Mid-Level Machine Learning Engineer to join our Model Engineering Team. This role is intended for engineers who view models as systems to be understood, developed, and improved rather than simply consumed as black-box components.
You will contribute across the full model lifecycle, from dataset preparation and architecture development to training, evaluation, and optimisation. Working closely with experienced engineers, you will improve model performance through experimentation, data-centric approaches, and systematic analysis, helping to build robust and efficient computer vision systems for real-world applications.
What You’ll Do:
- Train, fine-tune, and evaluate computer vision models for tasks such as classification, object detection, segmentation, pose estimation, and action recognition.
- Contribute to model development through architecture enhancements, custom heads, loss functions, and training strategy optimisation.
- Design and execute experiments to evaluate model improvements and performance trade-offs.
- Prototype, benchmark, and assess new architectures, techniques, and approaches against established baselines.
- Investigate edge cases and performance limitations, and improve model accuracy, robustness, and efficiency.
- Consider practical constraints such as latency, throughput, resource efficiency, and hardware-specific optimisation.
- Collaborate with engineering teams to improve dataset quality, model performance, and system reliability.
Requirements:
- Bachelor’s degree in Computer Science, Artificial Intelligence, Machine Learning, Data Science, or a related technical field.
- At least 2 years of hands-on experience training, fine-tuning, and evaluating computer vision models through academic, project, internship, or professional work.
- Hands-on experience with PyTorch and/or TensorFlow.
- Proficiency in Python and familiarity with Linux environments and command-line tools.
Bonus Skills
- Experience modifying models beyond high-level APIs (e.g. custom heads, backbone modifications, loss functions, or training pipelines).
- Exposure to multimodal, vision-language, foundation, or open-vocabulary models.
- Experience with optimisation techniques such as quantisation, pruning, distillation.
- Familiarity with experiment tracking, reproducibility, or MLOps practices.
Personal Competencies
- Strong interest in understanding how model architectures, data quality, and training strategies influence performance.
- Enjoy experimentation, performance analysis, and iterative model improvement.
- Eager to take ownership of model components and grow into increasingly complex model development responsibilities.
Note: Due to the nature of the role, the position is considered only for candidates who are already based in Singapore.