The Role

As an AI Engineer, you’ll join our core Deep Learning team and contribute across the pretraining and fine-tuning stacks, driving product-ready improvements in medical AI models.

You’ll work in a fast-paced, research-meets-product environment with high standards for scientific rigor and production-grade code.

What You’ll Work On

Pretraining (images ± text)

• Design and scale visual representations from large de-identified 2D/3D datasets

• Objectives: masked image modeling, self-supervised/contrastive (MAE/DINO), vision–language alignment (CLIP-style)

• Modalities: X-ray, CT, occasional MRI; 2D/3D ViTs and UNet-style decoders

• Infra: high-throughput DICOM loaders, augmentations, mixed-precision, distributed training

• Evaluation: transfer tasks, label efficiency, robustness across sites/vendors

Fine-tuning (Product Models & VLMs)

• Adapt and optimize models powering Gleamer’s clinical features

• Tasks: detection, segmentation, follow-up (temporal matching, tracking), calibration

• Vision–Language Models (VLMs): fine-tune encoders + LLMs for report generation and extraction

• Techniques: SFT, LoRA/PEFT, distillation, quantization

• Deployment: Docker, ONNX, TensorRT, batching, inference optimization

• Evaluation: AUROC/FROC, Dice, ECE calibration, factuality vs labels, clinician review

Tech Stack

• PyCharm, Windsurf, GitHub Copilot

• PyTorch (Lightning), MONAI, Hugging Face, timm

• ClearML, DVC

• Multi-GPU training, ONNX/TensorRT for deployment

• Clean, production-level code: type annotations, comprehensive tests, documentation

Must-Haves

• Strong ML foundations (probability, linear algebra, optimization)

• Proficient in Python + PyTorch; experience training/debugging deep nets

• Strong communication and scientific rigor

• Motivation to make a real impact in healthcare

Nice-to-Have

• Experience with self-supervised learning and/or VLMs

• Experience with medical imaging (X-ray, CT, MRI)

• Knowledge in 3D vision, uncertainty, domain adaptation

• Contributions to OSS or academic publications

1. Fit interview – informal conversation about your background and goals

2. Technical interviews – (Deep Learning Knowledge, Deep Learning Solution Design, Software Engineering)

3. HR interview – alignment on values, work style, logistics