Developing non-invasive brain-machine interfaces capable of predicting imagined movement with deep learning

• Implemented deep convolutional and recurrent neural networks to predict imagined directional movements

• Preprocessed EEG signal data to reduce noise and improve accuracy

Creating a web tool to identify AWS service failures and decrease remediation time.

• Developed a REST API to aggregate 10,000 service logs per minute from 20 AWS regions spanning 6 continents

• Designed and implemented a serverless backend with AWS API Gateway, Lambda, and Node.js.

• Created an interactive frontend with Typescript, React, and Redux to display tabular information.

• Integrated AWS CloudFront to reduce latency, Cognito to authenticate users, and CloudFormation to programmatically deploy infrastructure.

Accelerated ONNX Runtime to maximize inference rates and minimize power consumption for computer vision applications on Qualcomm mobile devices.

• Implemented an execution provider to perform inferencing on ONNX machine learning models

• Quantized deep learning models to reduce inference latency by 60x for deep convolutional neural networks such as MobileNet and ResNet