Christopher Forsythe Email & Phone Number

San Francisco Bay Area

• Currently working on defining the technical direction of the Biz Owner Discovery and Education team

San Francisco Bay Area

• Lead a large initiative to reduce workload of on-call engineer by coordinating multiple projects to improve observability, maintainability, and reliability of the notifications platform

San Francisco Bay Area

• Replaced the notification queuing system of Mercury (Yelps notification platform) to reduce costs and improve maintainability, observability, and scalability while also making Mercury self-healing• Configured SLOs (Service Level Objectives) for all services in the Mercury ecosystem to reduce on-call load and improve observability• Added multiple notification channels to Mercury (SMS, Email)• Improved push notification CTR (Click Through Rate) by 5% with implementation of intelligent push copy selection• Designed and implemented an email testing tool to enable Mercury clients to quickly test their emails

Monterey, California

Improved ​CylcWeb​ client load time by 40% and reduced data transfer from Django web server over 95% by implementing partial page updating.Reduced overall bandwidth load during working hours ~30% by engineering a flask server to bundle and compress files to be synced at a later time.Decreased deployment time of servers by 90% by assembling docker containers for automatic deployment

Monterey, California

Language Line Solutions does live audio and video language interpretation and is used commonly in emergency services to help those that struggle speaking EnglishI worked with a team to improve the accuracy and effectiveness of a trouble ticketing system. We planned to use a text classification model and train it using a website with millions of support ticket records. For the duration of the project I collaborated with the team to ensure a simple UX and functional solution.I designed a neural network to better predict customer call volumes for improved budgeting and planning. We devised multiple machine learning algorithms for text classification and time-series forecasting and utilized customer call patterns based on 5 years' of historical data while taking seasonality, holidays, and other factors into account. This was used to predict 2017 call patterns within 3% of previous manual effort.I designed and created a website for supervisors to integrate with machine learning model built for classifying free-text. This model surpassed accuracy of user-selected categories with minimal training.-Implemented Keras and scikit-learn to develop multiple machine learning models for time series prediction and text classification-Transmitted training data into Elasticsearch, and created Kibana visualizations and dashboards to provide analytics- Integrated Flask web server with corporate Active Directory(LDAP) for authentication and authorization.-Created HTML/CSS based on provided mockups with minimal guidance