Audrey B. Email & Phone Number

Burbank, California, United States

Burbank, Ca, Us

Burbank, Ca, Us

I specialize in AI, machine learning, and computational vision for StudioLAB, which supports the various studios within the WDS umbrella. I turn academic research into maintainable tools that support content production, post-production, and promotional materials.

Troy, Ny, Us

Troy, Ny, Us

Los Angeles, California, Us

Proposed and developed a novel method for embedding actions in images and videos by decomposing actions into subject, verb, object triplets. My method emulates linguistic constructs, allowing us to directly compare visual features and semantically-significant word embeddings.

Clifton Park, New York, Us

Researched, designed, and implemented a specialized object detector using Pytorch and other open-source machine-learning libraries. My research focused on implementing an adaptation of YOLO to improve performance on large-scale satellite imagery. This work formed the backbone of a system designed to correct imprecise human-annotated object detection.

Seattle, Washington, Us

As an intern with the Data Science team, my duties were twofold: prototyping and data analytics. With another intern, I designed and implemented a state-of-the-art, data-driven solution to assist in expediting accurate telemedicine diagnoses. I also performed data analytics and aggregation to assist Clinicians and Product Managers in analyzing patient demographic and medical information. This work introduced me to Keras, SQL, DataGrip, Amazon Redshift, and working with small data.

Seattle, Wa, Us

Working under Dr. Agnieszka Miguel at Seattle University, I researched various computer vision tasks, including motion analysis and object recognition. The goal of our research was to identify snow leopards in sets of wildlife camera trap images. We developed a software package that processed thousands of images by locating motion, marking instances of snow leopards, and identifying specific cats. This research directly supports the work of conservation biologists who monitor and preserve snow leopard populations across Central and East Asia.

I was selected to participate in the 2016 AMALTHEA NSF REU program. AMALTHEA stands for Advances in MAchine Learning in THEory and Applications, and was funded by the National Science Foundation's Research Experience for Undergraduates program, taking place over ten weeks at the Florida Institute of Technology.Our team was tasked with classifying infrasonic signals. We began by performing a week-long intensive literature review (which continued in a smaller capacity throughout the following nine weeks). After we had a firm grasp of the problem, we began by parsing and pre-processing the data in MATLAB, which included downsampling, filtering, normalization, and event isolation. We then moved on to feature extraction, where we employed two distinct methods, Mel-Frequency Cepstral Coefficients (MFCC) and the Hilbert-Huang Transform (HHT). Our intent was to compare the relative effectiveness of these two very different feature extraction methods.After extracting a series of feature vectors, we classified each signal with three different methods: individualistic k-nearest neighbors, Fisher's linear discriminant + mean-based kNN, and support vector machines. Our results were surprising and exciting, and we hope to publish our findings this year!

Kirkland, Wa, Us

Working for the Vehicle Electronic Control Unit Hardware team, I researched various functionalities of our trucks by studying technical documents and interacting with various other teams and departments, and used that knowledge to create hardware diagrams which greatly accelerated the design process of new electrical systems. I also worked with the Software and Vehicle Integration teams in collecting undocumented information for designing system diagrams. During my time with Kenworth, I created the primary electrical architecture document defining Economy Mode, an engine efficiency arbitration function. I also created over twenty hardware diagrams and eliminated the need for over forty others based on knowledge gleaned from studying various technical documents and collaborating with engineers in other teams, departments, and divisions.

Master Of Science - Ms, Computer Science

Bachelor Of Science (Bs), Electrical Engineering And Computer Science

Education record

Education record