Yinuo "Eva" Hu

(May 2022 - Present)

• Created an AI-based email topic detector using PyTorch, Google APIs and Flask that allows users to organize high quantity emails into subgroups, overseeing 3 data scientists and 2 SWEs
• Diagnosed and resolved data processing and data labeling challenges to improve the classification model accuracy by 30%
• Settled the B2B model for the company and defined the scope of the MVP after market research and competitor analysis

(Aug 2021 - Present)

• Created the pilot AI-based, personalized Student Evaluations of Teaching (SET) reports system to help instructors interpret SET results more efficiently and with less bias, and improved user satisfaction rating by 50%
• Developed the integrated AI modules using sentiment analysis, topic modeling, and extractive summarization, achieved SOTA model accuracies, and designed debiased algorithms to reduce bias by 40%
• Developed a web-based customer data platform prototype using SQL, JavaScript, Node JS, Express, and connected to Azure Cosmo DB.
• Built a visuzlied web app using React, Tableau, and D3.js, which will be used by over 13k instructors at the university

(Aug 2019 - Sep 2021)

• Led the web design team to build a lab website from scratch using JavaScript, HTML & CSS to communicate lab outputs
• Conducted several independent research projects using video data from the lab to conduct qualitative analysis and semantic network analysis with R and KBDeX, and used Tableau and Gephi to visualize figures
• Published and presented research findings at EARLI 2021

(Mar 2021 - Jun 2021)

• Proposed cost-saving strategies by replacing costly Azure API calls with OpenCV services - reduced cost by around 10%, and created demos of MVP solutions for client’s digital transformation.
• Scrapped warehouse-related information of over 100 companies using Python, organized the results in excel sheets, and proposed and pitched IoT & digital twin solutions to 3+ clients
• Developed a web-based customer data platform prototype using SQL, JavaScript, Node JS, Express, and connected to Azure Cosmo DB.
• Analyzed unstructured data to draft industry reports on elevating AI solutions to promote sustainability in Chinese and global market separately and presented the reports to 10+ stakeholders
• Outlined product vision and actionable strategy for next generation remote working software in Avanade Innovation Contest and pitched the solution to 50+ employees

SETSum is an AI-based, interactive website which aims to summarize and visualize personalized Student Evaluations of Teaching (SETs) reports for university instructors. The system incorporates three Natural Language Processing Modules:

• Sentiment Analysis
• Aspect Extraction
• Aspect-based Summarization

Hu, Y., Ryoo, K., & Tao, D. (2021). Scaffolding Collaborative Discourse and Modeling Practices. Poster presented at the Biennial Conference of the European Association for Research on Learning and Instructions (EARLI 2021), online.

Hu, Y., Zhang S., Sathy V., Panter A., Bansal M. (2022). SETSum: Summarization and Visualization of Student Evaluations of Teaching. Proceedings of the 2022 Annual Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies: Demonstrations (NAACL 2022) , Seattle, Washington.