• Independently created NLP model to detect dangerous speech, using HuggingFace open source pretrained models, and novel data labeling techniques based in best practices in psychology research. Created and managed active learning pipeline, and managed scaling with AWS Sagemaker, Cloudwatch, and EC2. Model is currently in production, and dramatically outperforming previously outsourced model from boutique NLP consulting firm.

• Developed full stack features related to data visualization, user management, and more, using React, Node, and AWS (dynamoDB, Lambda, S3, and more)

• Conducted technical interviews and assisted in hiring junior engineers

Created ML pipeline using AWS Sagemaker from preprocessing to training to deploying. Used Tensorflow Keras to deploy and modify semantic segmentation architectures (UNET and Deeplab). Managed AWS data pipelines with S3. Prepared reports for investors on ML training progress. Created robust evaluation notebooks and custom Cloudwatch metrics to track model metrics over time and training sessions.

Created company flagship mobile app singlehandedly from scratch using React Native with Redux for frontend, and Node.js for backend. Launched in Apple App Store and Google Play Store. Created extensive documentation and led training in React Native for other software developers to ensure a sustainable development workflow after contract was over.