SOFTWARE ENGINEER (MACHINE LEARNING), BSQUARE CORPORATION, WA. [MAY 2017 – PRESENT]
• Designed and led the architecture for next generation IoT Platform capable of multitenancy in AWS.
• Built a system to crop, rotate and resize images of chocolate bars using OpenCV and classify images using deep learning with Tensorflow and deployed in cloud and IoT Edge(vending machines) (Accuracy:88%). Optimized the model utilizing image histograms and custom one-class SVM algorithm to reach a cross validation accuracy of 95.07%.
• Built a real-time data streaming pipeline and state model workflow engine using business rules engine and reduced the end-to-end latency from 12 seconds to 400ms.
• Designed and developed serverless ETL pipeline using AWS lambda, Logstash and Elasticsearch for handling data, and customized Grafana to visualize it in near real-time.
• Developed and deployed machine learning pipeline in both serverless architecture and microservices with secure REST API interfaces using AWS lambda, ECS, Cognito.
• Developed a big data pipeline tool using Spark/Scala and RESTful microservices to do preprocessing and data augmentation.