Machine Learning / Computer Vision Engineer

25 Июля

Партнерские Вакансии

Компания "devs"

About the Role

We are a California-based product company. We’re looking for a hands-on ML/CV Engineer to design, train, and develop CV/LLM models that run at scale in production.

Key Responsibilities

• Build and refine algorithms for image recognition, segmentation, object detection, tracking, and classification.

• Train deep-learning models (CNNs, Transformers, diffusion models) in PyTorch / TensorFlow; run experiments, analyse metrics, perform hyper-parameter tuning.

• Prototype quickly, then turn research code into production-ready pipelines (Docker, CI/CD, cloud GPU jobs).

• Optimise inference speed and memory footprint (ONNX, TensorRT, model pruning/quantisation, CUDA kernels).

• Improve and maintain our existing CV services; monitor performance in the field and iterate.

• Collaborate closely with data engineers, MLOps, backend, and product teams throughout the full lifecycle—from dataset design to deployment and A/B testing.

Required Qualifications

• 2+ years of professional experience in Computer Vision or Deep-Learning R&D.

• Strong Python skills and fluency with core libraries: OpenCV, NumPy, SciPy, scikit-image, plus either PyTorch or TensorFlow/Keras.

• Solid grasp of ML fundamentals: optimisation, loss functions, data augmentation, transfer-learning, fine-tuning.

• Experience with version control (Git) and collaborative workflows (PRs, code reviews).

• Hands-on data handling: labelling, cleaning, balancing, augmentation, and dataset versioning.

• Ability to read & write technical documentation in English (B2+ or higher).

Nice-to-Have

• Familiarity with MLOps tools (MLflow, Weights & Biases, DVC, Kubeflow).

• Background in adjacent fields (3D vision, multimodal models, generative AI, LLM tuning).

What We Offer

• Competitive salary pegged to experience and market rates (rate review depends on your progress).

• Remote-first culture with a flexible schedule (sync hours overlap with California-time team).

• Hardware or cloud-GPU budget for experimentation.

*ДЛЯ ЗАКЛЮЧЕНИЯ КОНТРАКТА НУЖНО ПРОЖИВАТЬ НА ТЕРРЕТОРИИ РК И ИМЕТЬ ИП (ИНДИВИДУАЛЬНЫЙ ПРЕДПРИНИМАТЕЛЬ)