Machine Learning Engineer

Responsibilities

• Train, fine-tune, and evaluate computer vision models including object detection and image quality on real-world camera imagery
• Own the model-serving pipeline, including packaging models into NVIDIA Triton ensembles and deploying TensorRT engines
• Build and curate datasets using tools like FiftyOne and Label Studio to improve model accuracy
• Design rigorous evaluation harnesses, including regression suites and A/B comparisons, to measure model changes
• Develop LLM/VLM and agentic capabilities to extend self-hosted stacks and integrate them into product workflows

Requirements

• Strong proficiency in Python and the modern ML stack, specifically PyTorch
• Practical experience in computer vision, including object detection and working with real-world image data
• Proven experience taking models from experimentation (Jupyter/notebooks) to production deployment
• A self-starter mindset with the ability to independently investigate and solve ambiguous accuracy problems
• A rigorous approach to evaluation, focusing on datasets, ground truth, and edge cases

Preferred Qualifications

• Experience with NVIDIA Triton Inference Server, TensorRT, or DALI
• Familiarity with dataset curation platforms like FiftyOne or Label Studio
• Experience with LLM/VLM fine-tuning (LoRA), RAG, or multimodal models
• Knowledge of MLOps practices, including experiment tracking and model registries
• Exposure to edge/IoT computing or resource-constrained inference

Benefits

• Fast-growing business environment
• Fantastic office located in Tallinn
• Innovative and down-to-earth company culture
• Stebby wellness benefit
• Additional vacation and health days

About the Company

VOSKER is a leading provider of surveillance solutions for remote-area monitoring. We design intelligent, autonomous, solar-powered, and cellular-connected surveillance systems for the world’s most demanding environments, providing peace of mind and greater knowledge of the world.