Cynapse Pte Ltd – Computer Vision Model Engineer Intern

Job Description

We are seeking a hands-on Computer Vision Model Engineering Intern with a strong interest in training deep learning models, refining architectures, and building real-world computer vision systems.

This role is ideal for builders who enjoy running experiments, analysing model behaviour, and driving measurable performance improvements. You will work closely with experienced model engineers on production-grade computer vision models, ranging from standard detectors to modern zero-shot and open-vocabulary architectures.

What You’ll Do

• Train & Tune Models
  Train, evaluate, and improve deep learning models across a wide range of computer vision tasks, including (but not limited to) classification, object detection, segmentation, and action recognition.
• Architecture Exploration
  Research and experiment with model architectures (CNNs, Transformers), training strategies, loss functions, data augmentation, and optimization techniques.
• Failure Analysis
  Analyse model failures (false positives/negatives, edge cases) and propose practical solutions through data refinement, architectural changes, or improved inference strategies.
• Data Engineering & Experimentation
  Work with large-scale datasets, including data cleaning, preprocessing, and versioning to support reproducible experiments.
• Pipeline & Tooling Support
  Contribute to training and evaluation pipelines by improving automation, reliability, and experimentation efficiency.

Who This Role Is For

• Someone who enjoy training models, tweaking architectures, and iterating through experiments.
• Problem solvers who like debugging why models fail and understanding root causes.
• Builders who prefer hands-on engineering over purely theoretical work.
• Pragmatists excited to work on models that deploy into production environments.

Requirements

• Pursuing or completed a Diploma, Bachelor’s, or Master’s in Computer Science, Artificial Intelligence, Machine Learning, Data Science, Mathematics, or a related field.
• Strong interest in deep learning model training and experimentation.
• Proficiency in Python.
• Experience with PyTorch or TensorFlow through coursework or projects.
• Analytical mindset with the ability to reason about model behaviour and failure cases.
• Comfortable working in a Linux environment (SSH, file management, basic shell scripting).

Bonus Skills

• Hands-on experience training computer vision models (e.g., CNNs, Transformers, YOLO).
• Familiarity with vision-language models (e.g., CLIP, zero-shot or open-vocabulary detection).
• Familiarity with ML engineering tools and workflows, such as Git, Docker, and experiment tracking systems (e.g., MLflow, Weights & Biases, DVC).
• Knowledge of computer vision libraries (e.g., OpenCV) and video processing tools (e.g., ffmpeg).
• Interest in generative AI, multi-modal learning, or real-time video analytics.

Internship Details

• Duration: Minimum 6 months (flexible).
• Commitment: At least 4 days per week (full-time preferred).
• Schedule: Flexible start and end dates to accommodate exams or personal schedules.
• Location: Due to the nature of the role, candidates must be based in Singapore or have relevant experience studying/working in Singapore.