Maayan Keshet

I was the architect and implementer of a heavy-duty recording infrastructure for a cryptocurrency hedge fund, involving the design, coding, and management of servers spread across multiple geographical locations.

The design was focused on low-latency (want to be the first to trade), resistance to DDoS counter-measures and high-availability.

MAIN CHALLENGES
• Cost-efficiency: From the get-go, the design was dealing with a rapidly increasing number of exchanges and instruments, recording every move performed on them. My design was able to record dozens of instruments for cents per hour.
• High-availability: Despite making our systems fault-proof, cryptocurrency exchanges are still a far cry away from ideal with respect to availability standards common in the financial industries. Novel technical and statistical methods were used to overcome these potential gaps in data availability.
• In-house usability: Different operations (e.g., backtesting, arbitrage, training) require very different representations of the same data. The system had to supply and update the data in various functional forms, in an efficient manner.

I was in charge of the architecture and implementation of the notification mechanism (e.g., security-related monitoring, size changes) on a massively distributed file-system. It was typical for the file system (directories and all) to be spread across dozens of servers.

It was a difficult problem to solve in a time-sensitive manner, resulting in various industry leaders artificially capping their own solutions (e.g., NetApp and EMC at the time had a hard limit on the number of levels one can subscribe to).

My architecture and implementation handled this gracefully and without any feature limits placed on our clients. This resulted in a patent granted to Dell.