As a Deep Learning Engineer, you will be part of a talented team designing and training state of the art deep learning algorithms to identify placement, presentation, pricing, and availability of products in retail stores across the globe
In this role you will lead various initiatives designing, developing, and training in-house deep learning systems to perform operations such as optical character recognition, image captioning, object segmentation, object recognition, and more
Team: Kyiv — San Francisco
Project: The platform that builds automation solutions for the retail industry. The world’s first fully autonomous shelf auditing and analytics solution; one that provides retailers unprecedented visibility and insight into the state of merchandise in their stores
Experience / Skills:
Must have:
5+ years of machine learning background, with hands-on experience in building real systems
5+ years of experience with Linux & Command Line background
3+ years of experience using or building synthetic image generation systems, data augmentation pipelines, and OCR/image caption systems
3+ years of experience in debugging and diagnosing performance problems with ML algorithms
Demonstrated mastery of state-of-the-art machine learning and deep learning algorithms, techniques and best practices
Proficient in at least one of the following: Tensorflow, Keras, PyTorch. Tensorboard knowledge is a plus
Proficiency with training and running deep learning models on GPUs (both commodity and otherwise)
Demonstrated understanding of recurrent neural networks (including LSTMs and GRUs)
Proficiency with attention models, text localization, Google Cloud Platform, AWS, and serverless is a plus
Ph.D. or M.S. preferred
Demonstrated fluency in Python, other languages are a plus
Upper-intermediate English
We offer:
Great team spirit
Modern technologies and management methods
Career opportunities
Flexible working schedule, paid vacations and sick leaves
Medical insurance
Regular parties
Great office location near subway Maidan Nezalezhnosti and Zoloti Vorota
Responsibilities:
Participate in planning and prioritizing, write functional specifications and lead design reviews for our character recognition and image caption algorithms
Create photorealistic synthetic training data for augmentation
Develop, test, tune, and deploy both CNN and RNN based deep learning systems across a wide variety of customers
Manage and curate real-world training datasets
Evaluate existing frameworks and methods for speed and accuracy performance improvements
Collaborate with other developers, quality engineers, product managers, and documentation writers