About me

Elisa Uhura

Developed observers in C# for security services that saved hundreds of hours in engineering time by reducing manual validations and catching regressions.
Extended my intership DSL to support string interpolation with extensive validation to match Logic Apps DSL.
Proposed and execute a project to reduce test flakiness, addressing a variety of race conditions and unnecessary delays, substantially improving our CI/CD experience.
Implemented patches to enable Data Factory to be build in Linux and MacOS, including on ARM machines, which reduce the build time by 20% when compared to Windows nodes.
Maintained deployment templates using ARM to manage our services lifecycle in Azure, including AKS/Kubernetes clusters, Redis Cache, Key Vault secret management, Managed Identities, and others.
Proactively identified and addressed high-impact live production in our Petabyte-Scale SaaS ETL tool, including cross region issues with extensive use of Kusto/Data Explorer.

Architected and implemented a new low-latency pricing engine using StreamBase.
The new engine achieved a 10% lower latency, introduced support for multiple currencies, financial operations, custom risk model, and concurrent clients.
Implemented a new connector in Java to enable SignalR services to reliably use the new engine, reducing the cost of Forex to all clients due to the lower required margins to close operations.
Designed and implemented new tooling in Python to automate the deployment of production resources, effectively eliminating incidents caused by manual touches.

Designed, validated and implemented Pipeline Return Value support for Data Factory, driving meetings with multiple managers and PMs to design the API extensions.
Implemented the backend changes in C#, including a parser to convert from my DSL to LogicApps expressions. Implemented UI changes using Angular and TypeScript.
Created a demo to show that ADF Pipelines (by themselves) are now Turing-complete by computing the Fibonacci sequence using the new feature.

Ported and optmized computer benchmarks to use OpenMP Task-Dependency based parallelism and OmpSs to evaluate hardware-accelerated Task-Dependency runtimes.
Conducted literature review on computer benchmarks for parallel programming to determine the the best set of benchmarks to evaluate the benchmark.
Developed internal tools in Rust, Lua, and C, to measure and automate the evaluation of computer benchmarks.

GPA 8.5/10 (US 3.7), Graduated with honors, ranked 6th/72.
Teaching Assistant for Intro to Computer Science and Parallel Programming.
Promoted, presented, and managed multiple events, including hackathons and lectured courses in web technologies.
Assistant in HPC Research, implemented advanced measuring techniques and tuned C-based benchmarks like PARSEC to task based paradigms. Wrote educational material about software benchmarking published at ERAD/RS.