Computer Vision Engineer - Ultrasound AI | PyTorch

In this hands-on engineering role, you will take ownership across the complete segmentation model lifecycle—from curated clinical data ingestion and model training to evaluation, GPU inference optimization, cloud container deployment, retraining, and direct client collaboration. You will serve as a key technical interface, explaining model decisions, experiment results, and performance tradeoffs to both technical partners and clinical stakeholders.

Key Responsibilities

Model Training, Evaluation & Data Pipelines

• Multi-Label Image Segmentation: Train and refine multi-label medical image segmentation models using PyTorch to analyze clinician-annotated medical imaging datasets and ultrasound video streams.
• Experimental ML: Design and execute experiments to address class imbalance, overlapping anatomical structures, small/thin targets, and temporal context in video frames.
• Evaluation Frameworks: Build and enhance robust evaluation pipelines tracking per-label, frame-level, and event-level metrics while validating models using patient-level test splits and clinical reviews.
• Reproducible Pipelines: Contribute to reproducible training and retraining workflows complete with experiment tracking, checkpointing, automated testing, and CI/CD pipelines.

Real-Time Inference Optimization & Cloud Deployment

• GPU & Latency Optimization: Optimize inference pipelines for real-time video processing, prioritizing GPU latency minimization and throughput.
• Performance Tuning: Implement mixed-precision training/inference, model compilation, batching, frame striding, and hardware acceleration techniques.
• Containerized Cloud Deployment: Package, containerize, and deploy production ML models using Docker and AWS-based cloud GPU infrastructure.

Client Communication & Cross-Functional Collaboration

• Client Engagement & Demos: Participate in regular client calls, technical reviews, live product demos, requirement scoping sessions, and progress updates.
• Technical Communication: Clearly articulate technical decisions, model metrics, experiment results, architectural tradeoffs, and risks to both technical and non-technical stakeholders.
• Cross-Functional Synergy: Collaborate directly with clinicians, medical annotators, infrastructure engineers, and external ML research partners to translate clinical feedback into actionable engineering tickets.

Requirements

Technical Mastery & Software Engineering Foundations

• Experience Track: 3+ years of professional experience in software engineering, Machine Learning, or Computer Vision.
• Programming & Frameworks: Strong Python programming skills paired with hands-on experience training Computer Vision or image segmentation models using PyTorch.
• ML Lifecycle Understanding: Core understanding of model training, validation, evaluation metrics, data loaders, mixed-precision training, checkpointing, and GPU training workflows.
• Data & Partitioning: Practical understanding of video and image datasets, adhering to strict data separation standards between training, validation, and patient-level test splits.
• Cloud & Container Infrastructure: Experience or familiarity with AWS environments, cloud GPU instances, S3 object storage, and Docker containerization.
• Software Engineering Rigor: Strong engineering fundamentals including version control (Git), unit/integration testing, reproducible configurations, and continuous integration (CI).
• Client & Verbal Communication: Excellent verbal and written communication skills with the confidence to lead client-facing meetings, ask probing technical questions, and proactively communicate blockers or risks.

Preferred Qualifications

• Direct experience with medical imaging, medical image segmentation (e.g., DICOM/NIfTI formats), or medical video annotation platforms.
• Hands-on experience optimizing real-time Computer Vision or live video streaming inference pipelines.
• Familiarity with multiple segmentation architectures/frameworks (e.g., U-Net, DeepLab, Segment Anything / SAM, Mask R-CNN).
• Experience managing multi-GPU training setups and distributed ML pipelines.
• Proven track record working directly with enterprise clients, product leadership, or external clinical stakeholders.
• Awareness of Software as a Medical Device (SaMD) regulatory practices, model cards, traceability, audit trails, or clinical acceptance protocols.

Employee Benefits & Perks

• Financial & Security: Provident Fund, Interest-Free Loan Facility, and Advance Salary options.
• Healthcare & Wellbeing: Comprehensive Medical Inpatient and Outpatient facilities for employees.
• Work-Life & Lifestyle: Paid Overtime, Subsidized Lunch and Dinner options daily, Gym Facility membership, and Sports Allowance.
• Culture: Regular employee engagement activities, team events, and continuous professional growth opportunities.