Introduction to Yandex's YT Platform YT is an open source platform that is scalable and fault tolerant, with various tools for working with big data. It supports GPUs, making it a good tool for machine learning, and has a built-in OTP store system. YT is built around tables, which are schematized and can be used for batch processing or building OTP pipelines.

Data Processing and Compute Engines in YT YT's data processing layer is managed by the YT scheduler, which operates in terms of resource pools. The system supports the mapreduce paradigm, as well as vanilla operations and higher order compute engines like YQL, CHYT, and SPYT. YQL is a declarative and strongly typed SQL-like language, while CHYT is an adaptation of ClickHouse to the YT ecosystem and SPYT is a set of Apache Spark adapters for reading and writing data within YT. The system has a wide range of SDKs and a web interface for users.