● Built smart energy capabilities including user onboarding setup and animated solar savings dashboard using Python, C++(Qt), and Javascript for 80k Brilliant control panels. Interfaced with the backend, utilizing weather API and energy consumption data to enable real-time solar savings and production efficiency monitoring for solar panel builders and users.

● Optimized loading performance of the bluetooth smart device diagnostic screen for large homes with 70+ devices by redesigning the data model architecture/layers with Qt proxy models.

● Established the entire regulatory domain WiFi configuration for Brilliant controls. Utilized C++ and Javascript to build its UI setup flow, and Asyncio python for the backend logic of the embedded application that interfaced with connman.

● Handled edge case errors for device startup. Implemented error notification pop-ups for various wifi connection issues and

configuration timeout issues to allow users resolve the errors.

● Participated in building an analytical logging system that generates logs around screen views and button clicks behaviors.

Utilized these logs on Mixpanel to monitor feature engagement and traceback bugs.

● Investigated and resolved Sentry errors and bugs caused by race conditions, complicated dependencies, etc. resulting in a ~1000 error reduction per month.

● Built a 3D interactive web visualization tool that combined node edge graph with heat map using React and Three.js. Enabled biological science researchers to find data patterns in a high dimensional multi-omics environment with a team of 2 engineers and designers.

● Conducted user research through 6 qualitative interviews; pitched and presented the product demo to industry leaders and researchers to gather feedback for product prototyping

● Developed Home Screen Shortcut Navigations for Brilliant’s embedded home control with QML module and C++ to increase efficiency on accessing specific rooms and smart home devices; features deployed on over 50k Brilliant controls.

● Refactored the home control’s error notification system and home menu data structures with object-oriented design to increase its capability of handling large number of error states and diverse home device data categories.

● Collaborated with product managers and technical leads to define Brilliant home control’s functionalities and UI

● Design test suites and worked with the QA team to test functionalities on real hardware

A platform for students to watch lecture recordings uploaded by instructors with synchronous transcriptions in different languages

● Designed website user interface and developed the front-end features including the user sign up and video playlist page with React. Worked around cookies manipulation and designing customized UI components suitable for video steaming applications.

● Collaborated with a team of 20 using Git source control to review codes, test, and deploy changes to the server

• Utilized R to clean over 400GBs of Zillow housing assets and transaction data; merged housing transaction data with census tract and school district geographic data; stored data in PostgreSQL database; used QGIS to analyze these geospatial data and delivered results to a legal team as evidences of a complaint regarding the 2015 Flint water pollution incident

• Employed machine learning models and neural network with Python to help the Illinois Home Weatherization Assistance Program predict the housing energy consumption; improved the previous models and reduced the least square error by 20%

• Applied LASSO feature selection and used machine learning models on Python to implement unbalanced multi-class classification on ethnicity; discovered potential racial discrimination toward renters in rental processes