Location:United Kingdom (Remote or Belfast, NI Office based)
Position Overview
We are seeking aSenior Machine Learning Researcherwithdeep expertise in computer visionandhands-on experiencedeveloping andtraining large scale models (cluster-basedordistributedtraining). Your primary focus will be on object sensitiveneural compressiontechniques forlow-bandwidthstreaming androbust videoapplications. In this role, you will have the autonomy topush the state of the art,with ample opportunities topublishin leading conferences and journals.
This position offers theflexibilityto workremotely anywhere in the United Kingdom,with periodic visits to ourBelfastoffice if desired.
Key Responsibilities
Neural Compression Research
- Conceptualize and prototypeneural compression architectures(e.g. autoencoders, generative models) to enable high-fidelity image/video streaming under strict bandwidth constraints.
- Evaluate and refine models based onreconstruction quality, bitrate,andreal-timeperformance metrics.
- Develop models that enable high fidelity server-side reconstruction of objects of interests like text, vehicles etc.
Computer Vision Model Development
- Build and test downstream vision applications on reconstructed streams like
object detection, classification, segmentation, scene text recognition etc.
- Conductend-to-end experiments,from data preprocessing to model deployment, ensuring solutions meet performance and reliability targets.
Cluster-Based Training & Optimization
- Train large-scale CV models onmulti-GPUordistributedsetups, optimizing hyperparameters and resource usage.
- Profile training pipelines for efficiency gains (e.g., mixed precision, caching strategies), ensuring timely experiment cycles.
Edge Collaboration & Deployment
- Partner withedge engineersto tailor model architectures for resource-limited devices (ARM, Jetson, etc.).
- Support integration ofon-device inferenceandadaptive streaminglogic, balancing accuracy, latency, and energy constraints.
Research & Publication
- Stay current withcutting-edgeML and CV advancements, proposing novel research directions aligned with our project roadmap.
- Publish and present findings intop-tierjournals/conferences (e.g., CVPR,
NeurIPS, ICCV), showcasing both scientific and practical contributions.
Team Collaboration & Leadership
- Work closely withproduct management, data engineering,andfellow researchersto drive innovation and ensure project alignment.
- Mentor junior team members, promoting knowledge-sharing and best practices in research methodologies and code reviews.
Qualifications
- PhDinComputer Science, Electrical Engineering,or a related field, with astrong specializationinMachine LearningorComputer Vision.
- Hands-on experiencetraining and deployinglarge CV models,including somecluster-basedormulti-GPUexperience.
- Solid background inneural compression, autoencoders,or related techniques; familiarity withreal-time streamingconstraints is a plus.
- Proficiency indeep learning frameworks(e.g., PyTorch, TensorFlow) andGPU-acceleratedcomputing.
- Track record ofpeer-reviewed publicationsdemonstrating innovative work in ML/CV.
- Strongcommunicationandcollaborationskills, with the ability to translate research findings into clear, actionable insights.
- Self-drivenand comfortable takinginitiativein a dynamic research environment.
What We Offer
- Competitive Salary
- Comprehensive Benefits:Private health insurance, employer-matched pension, generous paid leave, and more.
- Work Flexibility:Fully remote (within the UK) or hybrid at ourBelfastoffice, supported by modern collaboration tools.
- High-Impact Research Environment:Access toadvancedGPU clusters and opportunities topublishin top-tier venues.
- Innovative Culture:Be part of a multidisciplinary team tackling challenging problems inneural compressionandcomputer vision,with real-world deployments.
Application Process
Interested candidates should submit the following tojanine@terminal-industries.com (via the via the‘Apply’ button above)
- ACV/Resumehighlighting relevant expertise.
- Alist of publications(or Google Scholar link).
- (Optional)Code samplesor links toopen-sourcecontributions.
Competitive Salary
