Software Engineer, AI/ML
At a glance
Mid-level AI Engineering role at Devsinc. Islamabad · full-time.
Pay not stated AI Engineering salary
aimlroles summary, based on the employer's posting.
What you'll do
- Train and refine multi-label image segmentation models in PyTorch using ultrasound footage and clinician-labeled scans
- Test difficult cases such as overlapping anatomy and small targets, then assess results by label, frame and event
- Check model quality with patient-separated test data and clinician feedback, then improve real-time video inference speed
- Package and deploy models with Docker and AWS, while maintaining repeatable training, retraining and testing workflows
- Work with the client and clinical and engineering partners to turn feedback into tasks and explain results and risks
What you bring
- At least two years in software engineering, machine learning or computer vision
- A bachelor's degree in computer science or artificial intelligence
- Strong Python skills and hands-on PyTorch experience, including training image segmentation or computer vision models
- Working knowledge of model training, validation, evaluation and inference, plus sound separation of training and test data
Who this fits
This suits a hands-on engineer interested in applying computer vision to clinical workflows and comfortable discussing technical choices with clients. The role is based on-site in Islamabad.
From the employer
We are looking for a Software Engineer with 2+ years of experience and a strong foundation in Machine Learning and Computer Vision to join a health tech project building real time AI assistance for clinicians performing ultrasound guided endoscopic procedures.
The system analyzes live ultrasound video and highlights anatomy, suspected pathology, and medical instruments frame by frame, helping clinicians understand where the AI model is focusing during a procedure.
This is a hands on engineering role where you will work across the segmentation model lifecycle, from curated clinical data and model training to evaluation, optimization, deployment, retraining, and direct collaboration with the client.
Benefits include Provident Fund, Medical Inpatient and Outpatient Facilities, Paid Overtime, Subsidized Lunch and Dinner, Gym Facility, Sports Allowance, Interest Free Loan Facility, Advance Salary options, and employee activities.
If you have hands on experience with Computer Vision and PyTorch, communicate confidently with clients, and are interested in applying AI to real time clinical workflows, we would love to hear from you.
Apply or reach out with your resume and a brief summary of your relevant Computer Vision, Medical Imaging, PyTorch, and client facing experience.
Requirements
Core Responsibilities
What you’ll work on:
• Train and improve multi label medical image segmentation models using PyTorch
• Work with ultrasound video datasets and clinician annotated medical imaging data
• Run experiments around class imbalance, overlapping anatomy, small or thin targets, and temporal context
• Build and improve evaluation pipelines with per label, frame level, and event level metrics
• Validate models using patient level test splits and clinician review
• Optimize inference for real time video processing with a focus on GPU latency and performance
• Work with mixed precision, model compilation, batching, frame stride, and other inference optimization techniques
• Package and deploy models using Docker and AWS based infrastructure
• Contribute to reproducible training and retraining pipelines with experiment tracking, checkpointing, CI, and automated testing
• Collaborate with clinicians, annotators, infrastructure engineers, and external ML partners
• Participate in regular client calls, technical discussions, demos, requirement discussions, and progress updates
• Explain technical decisions, experiment results, risks, and tradeoffs clearly to both technical and non technical stakeholders
• Work directly with the client to understand feedback, clarify requirements, and translate discussions into actionable engineering tasks
Qualification
Minimum: Bachelors in Computer Science/Artificial Intelligence
Required Skills
What we’re looking for:
• 2+ years of relevant software engineering, Machine Learning, or Computer Vision experience
• Strong Python programming skills
• Hands on experience with PyTorch
• Experience training image segmentation or Computer Vision models
• Understanding of model training, validation, evaluation, and inference workflows
• Familiarity with data loaders, mixed precision training, checkpointing, and GPU based training
• Understanding of video or image based datasets and good practices around train and test data separation
• Experience or familiarity with AWS, cloud GPU environments, object storage, and Docker
• Understanding of production ML inference and performance optimization
• Strong software engineering fundamentals including testing, version control, reproducible configurations, and CI
• Strong verbal and written communication skills
• Confidence participating in client meetings and technical discussions
• Ability to explain technical topics clearly, ask the right questions, and communicate blockers or risks proactively
• Ability to discuss experiments, model results, implementation decisions, and tradeoffs directly with the client
Nice to have:
• Experience with medical imaging or medical image segmentation
• Experience with medical video annotation platforms
• Real time Computer Vision or video inference experience
• Familiarity with multiple segmentation architectures or frameworks
• Experience working with multi GPU training
• Knowledge of model deployment and inference optimization
• Previous experience working directly with clients, product teams, or external stakeholders
• Awareness of Software as a Medical Device practices, model traceability, model cards, audit trails, or clinical acceptance processes