Machine Learning Engineer

Stack

Responsibilities

• ▹Build and improve ML models, mostly for computer vision tasks
• ▹Handle the full ML process: collect and prepare data, train and test models, improve them step by step
• ▹Work on standard and novel problems such as segmentation, detection, classification, diffusion models, LLMs and VLMs
• ▹Create fast prototypes and help turn successful ones into real products
• ▹Work closely with other teams to bring models into the product

Requirements

• ▹4+ years of real-world ML experience, including at least 2 years in computer vision
• ▹Good understanding of how to build ML systems from start to finish
• ▹Hands-on experience with one or more of: object detection, segmentation, img2img, generative networks
• ▹Ability to read, understand and implement ideas from cutting-edge research papers; stays current with top conferences (CVPR, NeurIPS, ICCV)
• ▹Strong Python and PyTorch skills
• ▹Experience optimizing models for efficient inference on local devices (CPU, GPU, NPU): quantization, pruning, runtime adaptation
• ▹Able to work with and explore large datasets
• ▹Comfortable working independently and taking full responsibility for tasks
• ▹Experience with prototyping and working in an R&D cycle

Nice to have

• ▹Research experience (PhD, papers or personal projects)
• ▹Training or fine-tuning foundation models (LLMs, VLMs)
• ▹Training or fine-tuning diffusion models for image generation or enhancement
• ▹Familiarity with classical image processing techniques
• ▹Experience with deep learning inference frameworks such as ONNX Runtime, OpenVINO, Core ML
• ▹Strong C++ skills, close to your Python level
• ▹Enjoy building and testing product ideas quickly

Soft skills

What we offer

• ▹Educational allowance
• ▹English and German classes
• ▹Choice of working remotely or in the office (Podil)
• ▹Medical insurance
• ▹21 days of paid sick leave per year
• ▹21 days of paid vacation per year
• ▹Flexible work schedule
• ▹Healthy fruit snacks
• ▹Frequent team-building activities

About the company

Skylum's software automates photo editing with AI while leaving all creative control to the artist. Its mission is to make photo editing enjoyable, easy and accessible to anyone.