1 месяц назад
Machine Learning Engineer (NLP/Computer Vision)
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Описание вакансии
Текст:
TL;DR
Machine Learning Engineer (NLP/Computer Vision): Building transformer-based data pipelines, production APIs, and machine learning systems for trademark search with an accent on NLP, computer vision, LLMs, and VLMs. Focus on fine-tuning pretrained models, designing scalable experimentation frameworks, and optimizing neural architectures in production.
Location: Helsinki, Finland; hybrid with 2 days in the office
Company
delivers AI-powered data, analytics, and professional services that help more than 5,000 customers protect and grow their brands.
What you will do
- Design and implement transformer-based pipelines for data processing and enhancement.
- Collaborate on NLP and multimodal machine learning tasks.
- Build and maintain production-grade APIs.
- Monitor, evaluate, and optimize models deployed in production.
- Apply, fine-tune, and develop pretrained LLMs and VLMs for trademark search and analysis.
- Design scalable data pipelines and experimentation frameworks while researching improved neural architectures.
Requirements
- End-to-end experience with NLP and/or computer vision pipelines.
- Strong software engineering skills with Python, Pytest, Docker, mypy, and algorithmic complexity.
- Proficiency with PyTorch, the Hugging Face ecosystem, and xTransformers.
- Deep knowledge of Transformers, RNNs, BLIP, LLMs, VLMs, and CLIP.
Nice to have
- Experience with LLaVA, Phi-3 Vision, Qwen-VL, or OpenCLIP.
- Familiarity with AWS IAM, SageMaker, EC2 Spot Instances, S3, and CloudWatch.
Culture & Benefits
- 9-month fixed-term maternity-cover contract.
- Part-time hours may be available.
- Hybrid work arrangement in Helsinki.
- Approximately 1,700 co-workers worldwide.
Будьте осторожны: если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого - это мошенники. Обязательно жмите "Пожаловаться" или пишите в поддержку. Подробнее в гайде →