Building GenAI for code generation and engineering efficiency.

Working on the largest deployment of Apache Spark in the world.

I worked on Resource Insights team that owns the data about all resources at Google (both physical and software ones). I work on the following projects:

• Distributed multithreaded framework for management and execution of data pipelines.

• Transformation pipelines that process more than 10 Pb of data.

• A system that collects data from 200 data sources.

• Anomaly Detection using Machine Learning techniques for multidimensional time series data that either comes as inputs to our system or produced by our system.

My main responsibility was the development of pricing and risk assessment platform for large portfolios of cross asset derivatives. I designed and developed the following:

• High-throughput (>100 Gb/h) scalable logging system for microservices architecture. Initial POC was based on Apache Kafka and Hbase. Production version was done with Apache Cassandra and Apache Spark. Web UI was developed with jQuery.

• Distributed microservices architecture with high-performance shared data storage based on Cassandra NoSQL database in order to speed up pricing and risk assessment

• General purpose aggregator for numerical results based on a long-running process in Apache Spark and data storage in Apache Cassandra

• Redesigned pricing engine for FX and commodity derivatives

Technologies: C++, Scala/Java, Python, JavaScript, Apache Cassandra, Apache Spark

• Researched and developed libtensor library for tensor operations in Q-Chem computational chemistry package (more than 3 million SLOC). Library has the ability to perform operations on both CPU and GPU and is implemented with C++ and CUDA C

• Build and administered two Linux clusters (25 and 6 machines) used for scientific calculations and software development (Linux, PBS, MPI, Kerberos, SVN, CVS)

• Performed an ab initio quantum computational research in the area of excited states