Computer Vision Engineer Offline
One of the subsystems of the platform is the Camera Area application that includes a recognition pipeline, monitoring components, backend microservices and mobile flatter-based applications. Internal communications between microservices are based on the Kafka. Data is persisted in a PostgreSQL database (Azure HDInsight).
The API Gateway is currently acting as an additional layer solely with the enterprise IT. The Gateway and the Backend are Java / Kotlin (Spring Boot) applications.
The backend consumes data from the recognition pipeline and provides data to the different Mobile Apps using Azure Push Notification as well as exposing endpoints. Mobile Apps are developed for Android and iOS. The recognition pipeline is based on the next tech stack: Nuclio, Python, Tensorflow etc. There is a monitoring service that leverages kafka streams and collects some business statistics that are published to Prometheus.
CI / CD are implemented using Azure Devops.
The recognition pipeline includes: Pallet Detection, Label Detection, Label Matching, Label Reading(OCR), Barcode detection, Barcode Reading, Delivery note identification, Analysis of packing structures and Data Validation.
The OCR based on the Google Vision API but developing of a custom OCR has been scheduled in scope of developing an embedded version of the recognition pipeline.
The pipeline communicates with the label management center, where clients can create and train matchers for new labels
*Responsibilities:*
- Investigating, prototyping and productizing solutions based on the computer vision technologies, deep machine learning models
- Find the optimal balance between setting and maintaining a research direction, and day-to-day delivery of robust technologies
- Determine implementation and optimization of algorithms
- Develop new and implement existing computer vision technologies for the logistic area
- Optimize code to perform in the Cloud. Edge Computing is a plus.
- Deploy models into our Cloud production environment that will be used by in a real-time setting
- Develop embedded computer vision solutions: android / iOs, microcomputers – e.g. nvidia jestson nano
- Develop ML pipelines for continuous machine learning (MLOps)
- Develop microservices and serverless functions based on computer vision technologies
- Develop mobile applications
*Requirements:*
- Experience taking a computer vision-based project from the initial concept to production
- Experience or strong familiarity with deep learning frameworks such as Tensorflow, Pytorch, Keras, Yolo etc.
- Experience training and serving computer vision models – object detection or segmentation, OCR, etc
- Solid knowledge of OpenCV
- Experience developing software and algorithms for computer vision, machine learning or computer graphics applications
- Deep understanding of machine learning fundamentals
- Experience with open source computer vision and linear algebra frameworks
- Strong instincts for efficiency and optimization
- 3-4 years coding experience in one of these languages: Python, C++, Java, Kotlin
- Highly competent in data science techniques and software libraries
- Experience with dockerisation of computer vision components
*Will be a plus:*
- Experience in Cloud based AI/ML platforms such as AWS Sagemaker, Azure ML, Google AI products
- Knowledge of Kotlin (Java)
- Understanding of microservice patterns
- Knowledge of Kafka and experience with scalable recognition pipelines
*We offer:*
- Flexible working hours
- A competitive salary and good compensation package
- Best hardware
- A masseur and a corporate doctor
- Healthcare & sport benefits
- An inspiring, comfy, clean and safe office
*Professional growth:*
- Challenging tasks and innovative projects
- Meetups and events for professional development
- An individual development plan
- Mentorship program
*Fun:*
- Corporate events and outstanding parties
- Exciting team buildings
- Memorable anniversary presents
- A fun zone where you can play video games, foosball, ping pong, and more