Smart call assistant

UHP, Deutsche Telekom
2 years

Smart Call Assistant is a voice-controlled AI-powered service that overcomes the drawbacks of synchronous telephone communication.

It provides features for intelligently answering calls in situations where you cannot speak but can still communicate.

Project role
Main backend developer
Responsibility
  • Leading the implementation of microservices that connect mobile app to ML service.
  • Making sure that the services are written following the best practices like TDD, SOLID and CLEAN.
  • Modeling the data layer.
  • Mentoring the colleagues.
  • Choosing the right technologies for our solutions.
  • Making sure that the product is well documented.
Technologies & Tools
Python 3.10 Python 3.8 RasaAI Starlette FastApi SqlAlchemy Alembic Asyncio PostgreSql GitLab Azure Open Telekom Cloud Grafana Kibana Kubernates Linux MS techs: TTS, STT Postman

Smart call assistant was the most complicated product I worked on, the only product that worked with real calls, mobile and landline, and used the technologies that are unknown for software engineers that never worked with Telekom.

Also, on the project we had international team with 20+ members, which was the biggest team I worked with and the only team where the whole communication was in English, over 2 years.

There was mobile team, backend (my team), ML team, voice specialists and others.

This project was the only project where we had on the backend 100% tests coverage, which I am really proud of.

It was the part of Magenta Voice program.