Wrote asynchronous C++ server to simultaneously handle data input and manage resources with low latency.

Designed Autoscaler server to be fault-tolerant and persistent on server crash with data-structure serialization.

Implemented cascade structure with memory-mapped allocator for fast and persistent timeseries storage.

Wrote web endpoints to query from and write to Postgres database in an efficient manner using indexes.

Talked with customers about their LLM inference server needs and designed features around them such as metrics endpoints and informative error handling. Wrote intuitive CLI, docs, and guides for Autoscaler.

Wrote concurrent and asynchronous python server to authenticate and benchmark requests to LLMs.

Researched and wrote Competitive Analysis and Market Map PowerPoint for new software product.

Conducted user interviews, wrote functional and architectural design, and implemented UI tool in MATLAB to facilitate Machine Learning model documentation. Used Jira for Agile development practices.

Implemented Deep Learning model for Computer Vision using Pytorch. Wrote scripts to unit test model, train model, and visualize model outputs, achieving state-of-the-art performance for keypoint detection.

Productionized recursive matrix decomposition algorithm and made tests to ensure accuracy and performance.

Designed Gaussian Process model for time series with training and inference script in Pytorch with tests.

Wrote scripts for Infrastructure as Code processes and CI/CD in a Docker-like container for a cloud service