Rustem Kamun

The main deliverable of my master's dissertation is a clock synchronization protocol (CSP) for general-purpose transactions at scale. In designing CSP, we employed a hybrid logical clock, HLC, that leverages the best of logical clocks and physical clocks. HLC is helpful for tracking causality relationships of the overlapping events. To maintain clock offsets among sets of nodes, we utilized Marzullo intersection algorithm. Synchronization of clocks in a network requires comparing them directly or indirectly with respect to differences in time and (possibly) frequency. The approach adopted here is based on variant of Christian's algorithm (Probabilistic clock synchronization).

Built and maintained an SDK for Python. This SDK provides a small engine and tools for translating Python-based applications into any language using the TranslationExchange.com service (https://github.com/translationexchange/tml-python).

Available for Flask/Django/Python

Developed an advanced tool for efficient analysis of large data sets with arbitrary structure. Designed scalable data storage using MongoDB. Building core that handles MongoDB constraints and limitations. Designed, developed, and tested efficient and CPU parallelizable data filtering algorithms. MVP was based on Django, RabbitMQ, Gevent, and MongoDB.

Building an ETL tool using Python Pyramid. Concurrent session management and data extraction and transform tool which is seamlessly compatible with SQL-family databases, including Redshift, Oracle, MySQL, PostgreSQL, and MS SQL.