Senior Research Engineer - Computer Vision at Instadeep

The Team:Our Computer Vision team builds and deploys cutting-edge models that power state-of-the-art perception and analysis across multiple domains. From biomedical imaging and robotics to satellite vision and digital twins, we develop algorithms and scalable pipelines that transform raw data into meaningful insight. Working closely with research and product teams, we translate foundational research into robust, high-impact applications.The Role:We seek an experienced and highly skilled Senior Research Engineer in Computer Vision for Biology with a passion for designing and optimising advanced visual intelligence systems that unlock insights from complex biological data. You’ll lead projects that combine deep learning, efficient architectures, and large-scale training to solve perception problems in microscopy, histopathology, and other biomedical imaging domains. If you love transforming the latest computer vision research into powerful tools for scientific discovery and translational biology, this role is for you.As a senior member of the Vision team, you will lead technical development, mentor junior engineers, and help shape the strategic direction of bio‑vision AI initiatives at InstaDeep.TL;DR: Develop and optimise next-generation biological vision systems — from foundation models for microscopy and whole-slide imaging to multimodal bioimaging pipelines — with maximum efficiency, scalability, and scientific impact.

Responsibilities

Model Development: Research, design, and implement state-of-the-art computer vision models (e.g. Vision Transformers, SAM, MambaVision, Segment Anything, Diffusion Models).

Algorithmic Optimisation: Translate the latest deep learning research into performant, production-ready systems.

Scalable Vision Systems: Build distributed training and inference pipelines optimised for large-scale datasets (e.g. WSI, medical, industrial, or satellite imagery).

Multi-Modal Learning: Integrate vision with other modalities (e.g. text, genomics, spatial omics, or language) to build cross-domain AI systems.

Performance Engineering: Profile training workloads, optimise GPU utilisation, and accelerate inference through efficient data pipelines, caching, and mixed-precision training.

Collaboration & Mentorship: Partner with ML researchers, engineers, and domain experts to deliver high-impact solutions and guide junior engineers.

Project Leadership & Client Engagement: Lead and manage end-to-end machine learning projects, coordinate cross-functional teams, and act as the primary technical point of contact with clients to ensure successful delivery and alignment with project goals.

Required Skills

Strong experience in computer vision and deep learning (classification, segmentation, detection, representation learning, or generative models).

Expertise with deep learning frameworks (PyTorch, JAX, TensorFlow).

Proficiency in Python and familiarity with C++ or CUDA for performance-critical components.

Experience with distributed training and large-scale data processing frameworks (e.g. Ray, PyTorch Lightning, Dask, or Horovod).

Solid understanding of visual data pipelines, pre-processing, augmentation, and evaluation metrics.

Experience deploying and optimising CV models on GPUs/TPUs.

Familiarity with Linux environments, Git, Docker, and cloud/HPC platforms.

Highly desirable:

Experience training or fine-tuning vision foundation models (CLIP, SAM, DINOv2, ViT, Mamba, Diffusion Models).

Understanding of 3D vision, multi-view geometry, or spatiotemporal data.

Experience with high-throughput image data systems (e.g. for medical or satellite imagery).

Knowledge of MLOps and efficient model serving frameworks (ONNX, TensorRT, TorchScript).

Strong publication or open-source contributions in computer vision research.

Example Projects:

Train and optimise foundation models for high-resolution image analysis (e.g. gigapixel WSIs, aerial imagery).

Develop active learning and segmentation systems for large-scale annotation workflows.

Integrate Mamba-based or transformer-based architectures for multimodal vision-language models.

Build efficient real-time inference pipelines using CUDA/XLA and distributed systems.

Deploy scalable visual search and embedding-based retrieval systems.

Apply Now