Hawthorne AI specializes in 2-6 month software projects in Machine Learning, Computer Vision, and IOT. We will scope and design your project, implement custom end-to-end machine learning pipelines on time, and offer post-delivery support and management.

Built Automatic ML Pipeline using AWS Sagemaker, consisting of 40+ GB of data cleaning/preparation, model training,
model optimization, multi-tenancy, and deployed event recommendations to 11 million users through our web and
email services
● Built systems with: Tensorflow, Keras, Pandas, Numpy, Flask, Docker, Nomad, CircleCI, AWS Sagemaker, ECR, ECS, S3,
Lambda, Cloudwatch, Redshift, Databricks

Implemented and deployed several GCP Cloud functions for user feedback/exp using Nodejs and Twilio API.
● Deployed and trained a conversational NLP agent webhook using Twilio/Stripe/Intercom/Dialogflow API’s in Nodejs that
understands and responds to user texts.
● Data Analytics: Created diagrams and charts, Reporting key business metrics such as active/inactive users, purchases per
user, avg purchase time, etc using SQL, Python, Pandas.

1.5x model accuracy by building a CV pipeline for training and evaluation that uses YOLOv3, Tensorflow with transfer learning, Darknet, and OpenCV to detect and track objects which enter dangerous areas on industrial job sites.
● Launched our core ML services with Tensorflow/Keras and AWS IOT running on Raspberry Pi’s, and streaming incident
videos to GCP.
○ Exclusion Zone Monitoring for People/Vehicles/Equipment, Model Evaluation, Personal Protec

Trained, implemented, and deployed state of the art CNN models to accurately show human 3D Pose Estimation in real
time using Tensorflow, Keras, MS Azure.
● Implemented a real time 3D hand tracking pipeline consisting of object detection, tracking, and 3d joint lifting, which
transfers webcam data into virtual reality, allowing our product to display a person’s hands in VR.

Acted as a Project Manager and performed data science research with a team that applied Deep Learning and
DeepMind’s EWC to a Fetch robot for Image Recognition with ROS, Tensorflow, Keras, and Python, achieving 96%
accuracy.
● Designed a probabilistic framework for short-term person recognition using Expectation Maximization, K-means, and
Deep Learning with Opencv and Tensorflow.

Designed and implemented a vehicle state machine model to make a modular system for cross-team validation of the
Model 3.
● Implemented a Python library and several test modules for NFC Locking, Drive-Away Locking, Theft Prevention, and
other requirements of the Model 3 Security Controller.

Designing electrical systems, 3-D modeling/printing, and making mechanical decisions for a network of Android tablets,
controlled by Raspberry Pi’s, which users can access remotely.
● Writing Android tablet stress tests in Python for an automated robot controlled by an Arduino.
● Validating and configuring Linux and Android systems and tablets according to developer requirements.

Developing automated stress and functional testing for Android tablets.