Sergey Parakhin

Sergey Parakhin

Sergey Parakhin joined Grid Dynamics in 2014. He has been a technical lead for projects focusing primarily on scalable distributed systems and machine learning (natural language processing and computer vision). Sergey holds a BS in Software Engineering from the Kharkiv National University of Radioelectronics.

Sergey Parakhin

Sergey Parakhin joined Grid Dynamics in 2014. He has been a technical lead for projects focusing primarily on scalable distributed systems and machine learning (natural language processing and computer vision). Sergey holds a BS in Software Engineering from the Kharkiv National University of Radioelectronics.

Sergey Parakhin

Building a question answering system for online store

The use of question answering (QA) systems is becoming more prevalent in various domains, including customer Q&A and reviews for products such as photo and video cameras. These systems can help users find information more efficiently and provide direct answers to their questions. Building and adapting QA systems to specific domains can be challenging, but techniques such as knowledge distillation and pre-training on unlabeled data can improve the performance of these systems. Additionally, optimizing the models for production, such as using GPUs for inference and applying pruning techniques, can help accelerate the models and make them more suitable for real-time applications.

How we built a conversational AI for ordering flowers over Amazon Alexa or Google Home

Grid Dynamics Labs has developed a voice assistant called “Flower Genie” for Amazon Alexa and Google Home platforms, allowing customers to select and order flower bouquets using voice-based interactions. The article discusses the process of building the conversational AI system for ordering flowers and the challenges faced in conversation design, data collection, natural language processing, testing, and certification.