Terrence Ho Email & Phone Number

New York, Ny, Us

Menlo Park, California, Us

Menlo Park, California, Us

Menlo Park, California, Us

Los Angeles, California, Us

Los Angeles, California, Us

Los Angeles, California, Us

Redmond, Washington, Us

• Developed tool to diff Azure VM security scans over time and reduce data needed to be analyzed by 10-50%.• Built service to automatically ingest new scans then compute and upload differences to Azure Log Analytics.• Scaled diffing to all Azure internal VMs by decreasing diffing time from 3 secs to 15 secs with parallelization.• Added execution timestamps and SHA1 hashes to security scans to identify programs used in a breach.

Palo Alto, Ca, Us

At the time of my internship, Stockpile was revamping their mobile app with Stockpile App 2.0. My job was to:1) Create a GraphQL orchestration server. Written in Node.js, it wraps Stockpile's existing backend API and performs request orchestration for the future mobile application, allowing Stockpile's apps to request and receive all it's data at once, rather than make many individual requests. This resulted in a large speedup when the new app was started. The GraphQL orchestration server was crucial to providing a frictionless user experience.2) Co-develop the next generation iOS app in native Swift. To increase startup performance, Stockpile decided to move away from React Native, a complete rewrite. I created the new Dashboard, complete with interactive stock graphs and tickers. Additionally, all models and data were fetched from the GraphQL orchestration server. I was able to bundle all startup requests into one request, drastically speeding up the new app's startup time. 3) Create end-to-end tests for GraphQL orchestration server, using Mocha. I integrated a continuous integration/continuous dvelopment cycle into the daily workflow with a custom configured Jenkins, automatically running tests and deploying the GraphQL server everytime a commit was made. For the new iOS app, I created UI tests using XCTests, and used Bitrise to automatically run the tests and generate new QA builds on every new commit.4) Fix bugs and smoothen user app flows in Stockpile's current React Native mobile application. For example, when a user first signed up, they should not be able to add money into their account until Stockpile verified their identity. However, user flows in the React Native app allowed users to add money before their account was ready. I closed this gap, and kept Stockpile in legal compliance with brokerage laws.

Westlake Village, California, Us

My project mainly focused on developing machine learning financial models. Given a set of mortgage loans that are in danger of defaulting, I created a model to predict which loans were most likely to experience a turn around and payoff the loan. Final models obtained 93% accuracy, compared to the previous models that tested around 80% or lower. To obtain this result, I combined an ensemble of learning methods, including Random Forests and Logistic Regression. This combination was selected to increase both precision and recall of our models.

Master Of Science - Ms, Computer Science

Bachelor Of Applied Science (B.A.Sc.), Computer Science