MEET ALLEN YE

---

Tesla Autopilot

Software engineer intern and incoming full-time software engineer for the AI Tooling Team

Amazon

Software developer intern at Alexa AI, local information team. Created a Java package to be consumed by client packages for an end-to-end data pipeline integration that captures customer feedback metrics and signals. Leveraged AWS services such as OpenSearch for creating indexes to store/retrieve data and performed data analysis and visualizations in Kibana.

Northrop Grumman

Used C++ to improve the Sustainment and Modification of Radar Sensors (SMORS) system for detecting aerial objects and threats. Added new capabilities on the radar operation system while working in an Agile environment. Led other interns within the team and coordinated tasks to efficiently resolve multipe medium and high priority tasks to enhance the GUI display of the Mission Application Software.

NASA

Worked on improving the efficiency of the National Airspace System by optimizing flight operations through analyzing and extracting semi or unstructured information from flight documents (Operations Plans). Conducted unsupervised learning experiments by clustering high dimensionality data using NLP methods (spacy, tf-idf, UMAP, DBSCAN). Also assisted in evaluating a speech to text NLP model specific to FAA terms for transcribing FAA webinar meetings

Stanford University

Evaluated the accuracy of a fast online linear algorithm in Matlab and created input data of various distributions. Used Matlab’s CVX and Mosek ApS solver to compare fast online linear algorithm effectiveness to an offline algorithm.

San José State University

Researched the applications of AI with ensemble learning for detecting malware and published the chapter “On Ensemble Learning” in the Springer textbook “Malware Analysis using Artificial Intelligence and Deep Learning”. Processed 80 GB of raw malware data, extracted opcode features using data pipeline, and trained various machine and deep learning models including CNNs (1-D and 2-D), SVMs, MLPs, KNNs, and ANNs. Applied ensemble methods of bagging, stacking, and boosting to complement previously mentioned models. Achieved balanced accuracy of 88.16%, precision score of 93.84%, recall score of 93.37%, and F1 score of 93.13%.