Senior Computer Vision Engineer – Cloud-Native Model Deployment

Harbor Softworks

Contract · Remote · Flexible hours

About the role

Harbor Softworks is strengthening the security posture of its cloud‑hosted services. As a Computer Vision Engineer you will lead the vision‑based threat detection module codenamed "Sentinel Lens" within this initiative. The project focuses on hardening image‑processing pipelines against adversarial attacks while maintaining high throughput.

Key responsibilities

• Design and implement robust CV models in PyTorch to detect malicious image uploads in the "Sentinel Lens" pipeline.
• Integrate model inference into AWS Lambda functions, ensuring end‑to‑end encryption and audit logging.
• Conduct adversarial robustness testing using FGSM and PGD attacks, and document mitigation strategies.
• Collaborate with the DevSecOps team to embed CV security checks into CI/CD pipelines via GitHub Actions.
• Produce performance benchmarks (latency, false‑positive rate) for each iteration and present weekly to stakeholders.

Must-have skills

• 5+ years of Computer Vision experience, primarily with PyTorch or TensorFlow.
• Strong proficiency in Python for model development and API integration.
• Hands‑on experience securing cloud‑native workloads on AWS (S3, Lambda, IAM).
• Knowledge of adversarial machine learning techniques and mitigation.
• Familiarity with containerization (Docker) and orchestration (Kubernetes) for model deployment.

Nice to have

• Experience with MLOps tools such as MLflow or Kubeflow.
• Background in cybersecurity or threat modeling for AI systems.