{
  "organization": "Microsoft",
  "title": "Software Engineer",
  "description": "Building Databases at Azure Core now :), but before that I designed and refactored control plane to data-plane data-seeding framework to reduce instance creation time from our RP and provide a common base for all data-plane seeding operations for resiliency on common transient scenarios, decreased end to end latency on resource creation, and decrease the amount of redirection our V1 introduced across system boundaries.\r\nImplemented “reference data seeding” for ADME data-plane. The challenge was to beat existing open-source code that seeds data into our Data plane from files on disk. I wrote a custom Concurrency throttler that resulted in an 85% efficiency gain. Tech stack was C#.  Ideated, and Built LogParsely a real time log ingestion tool that takes a process’s standard output that returns JSON, ingests it in a flattened format regardless of nesting, and lets you write SQL queries against ingested logs. This tool was written in Rust, and is now used by the ADME engineering team for their inner loop testing. The tool also supports resiliency against common disk IO errors and is built into our Developer Environment by default now. Implemented the direct data upload and retrieval POC for Seismic Data Management Service on Azure Data Manager for Energy. This involved efficiently handling multipart file streaming on the node based Seismic Service, and benchmarking performance. Decreased Bandwidth throttling by 30% Tech stack: C++, Node, Python, and created the internal data plane management framework, which leveraged existing architecture, and lets developers write workflows that could be executed by a common api and infrastructure for running maintenance jobs on data plane instances. Tech stack was C# and dotnet 6",
  "started_at": "August 2022",
  "ended_at": "Current",
  "type_of": "Job"
}