CV
Please kindly find my Curriculum Vitae.
Education History
- B.S. in Computer Science, University of Notre Dame
- Aug. 2016 - May. 2020, GPA: 3.84/4.0
- SCPD NDO, AI track, Stanford University
- 2020 - Present, GPA: 3.94/4.0
Work experience
- Software Engineer, Google, Mountain View, CA (Sep 2022 - Present)
- Advertisement Campaign Platform
- Software Engineer, Repo Ticketing and Lifecycle, Bloomberg, New York, NY (Dec 2020 - Sep 2022)
- Refactored the most heavily used ticketing screen shared across multiple teams; broke down the monolith repository (20k lines of code) consisting of several team’s business logic into submodules; reduced bug-related tickets by 30%, while getting a more extendable, testable codebase that significantly increased productivity.
- Acted as the engineering lead of a redesign of an overnight carry-over rate calculation API for Repo (repurchase agreement) ticket; engaged with clients to clarify the functionality of the API; dissected the project into deliverable stories and prioritized based on Agile development method; followed through the lifecycle of the project development after launching to production.
- Led the revamping of financial figures calculation of Repo ticket by leveraging an internal third-party API; hosted several design meetings within the team and documented project progress; engaged with API owner for question clarifications; prototyped a design to prove the feasibility.
- Observed and resolved the slowness issue during manual ticket booking by adding RabbitMQ to the ticketing service to honor manual ticket booking over ticket booking in the background; greatly improved the manual ticket booking speed.
- Mentored a new developer of the team during my second year at Bloomberg
- Software Engineer, Data Intelligence, Bloomberg, New York, NY (Jul 2020 - Dec 2022)
- Built a new internal machine learning framework handling the entire development pipeline from data serialization to model deployment, facilitating more independent model development and precise business responsibility dissection.
- Completed document classification and entity extraction projects from financial documents, using tools such as GiGwork and Snorkel for data collection and augmentation, PyTorch and sklearn for model development, and Kubernetes for automatic model deployment.
- Software Engineer Intern, Bloomberg, New York, NY (May 2019 - Aug 2019)
- Enabled memory-sharing among ticker message queues with atomic lock-free queue; allowed less active tickers to be grouped together and freed up memory resources to accommodate more clients, leading to 10X boost in resource utilization efficiency.
Research experience
- Data Mining towards Decision Making Laboratory, Notre Dame, IN (Sep 2019 - May 2020)
- Systematically designed, prototyped and characterized an array of natural-language generation (NLG) models for producing a target sequence (i.e., a news story) based on a given list of entities
- Studied the impact of using different levels of input entity representations (name, mention, type), models, and mechanisms on text generation quality.
- Developed a novel type-guided generation model within sequence-to-sequence (Seq2Seq) learning framework which incorporates entity type information during decoding as a guide for more accurate context words generation. Our proposed model outperformed the baseline methods including pre-trained models such as UniLM by 5%
- Notre Dame Social Sensing Laboratory, Notre Dame, IN (Aug 2018 - Oct 2019)
- Developed PQA-CNN, a perceptual quality-assured convolutional neural network framework, to solve the single-image super-resolution (SISR) problem in remote sensing, i.e., reconstructing a high-quality satellite image from a low-quality one (the paper was published at IWQoS’20, best demo award).
- Developed GraphCast, a multimodal graph neural network framework, to predict the urban traffic risks at a fine-grained spatial scale by jointly exploiting the data from social media sensing and remote sensing (the paper was presented at SECON 2020).
- Developed SyntaxLoc, a syntax-based probabilistic learning framework using social sensing to extract the location entities from social media context, focusing on addressing the challenges associated with social media posts’ limited and unstructured content (the paper was presented in the ASONAM’19).
- Personal Robot Group @ MIT Media Laboratory, MA (Jun 2018 - Aug 2018)
- Developed an integrated workflow to enable a social robot learning companion (used in education as a tutor or a peer learner) to perform active role adaptation based on users’ feedback
Extracurricular experience
- Alto Saxophonist, Notre Dame Marching Band, University Band, IN
- Delivered performance on behalf of Notre Dame in an array of on-campus and off-campus events, e.g., the annual performance at DeBartolo Performing Arts Center.