Cornelis Jan Drost

This is a local scripted workflow and AWS-hosted management system for the assignment, prioritization, and tracking of data analysis tasks to be completed by a team of over 150 analysts.

Tasks spanned a multi-step workflow, such that the completion of one task could trigger the queueing of multiple dependant tasks or that the failure of a QC step could cause a task to be repeated. A key feature of the system was prioritization, allowing managers to set priority rules based on task age, type, and metadata retrieved from other sources. The ability to do this at a high level saved over 50 hours a week across six managers compared to the previous manual workflow. The local workflow streamlined or automated away almost all manual data management, eliminating most opportunities for human error and allowing analysts to progress cleanly between tasks.

I designed, developed, and supported the local workflow component of this system, which was in successful use for over eight months while the online component was developed. I also designed, developed, and launched the online component while overseeing the analyst team's day-to-day support and feature requests.

This Python software development kit (SDK) supports the development of smart contract code in the Etch language for the Fetch.ai blockchain.

My primary contribution was the development of a lexer and parser extension for the Etch language. This allowed basic syntax checking before a contract was submitted and allowed the extraction of configuration details from the code. For example, Etch contracts could declare global variables, which would persist on the blockchain, either across all shards (i.e., parallel executions of the chain) or a subset. Correct deployment of contracts containing globals required matching the shards used by the code to those to be deployed to.

I also oversaw keeping the SDK up to date with the rapidly developing ledger API.