NIAN LIU | JOB HUNTING
WORK EXPERIENCE
Zilliz San Francisco / Shanghai, Hybrid 2023.6 -
Software Engineer
Cloud | Gen AI
-
I was part of the team responsible for developing the Zilliz Cloud platform, a leading vector database company best known for Milvus. Our primary goal was to expand the availability of Zilliz Cloud on major cloud providers such as AWS, Azure, and GCP across multiple countries. This expansion aimed to establish a robust long-term memory foundation for Large Language Models (LLM).
-
In addition, I made contributions to Gen AI projects related to vector databases, including RAG and data pipelines. These projects were focused on enabling organizations to rapidly develop AI/ML applications and harness the potential of unstructured data.
Amazon Web Services Vancouver, Canada 2021.12 - 2023.3
Software Engineer
S3 Index
- I worked with the S3 Index at AWS, where I was part of a team focused on developing Incremental Backup Services. Our mission was to provide robust solutions for backing up the S3 Index, including critical data and partition metadata, to ensuring that the index data, which is fundamental to the performance of the service, could be securely and efficiently backed up.
Concordia University Montreal, Canada 2019.9 - 2021.9
Research Assistant
Characterizing Deprecated Deep Learning Python API: An Empirical Study on TensorFlow (Paper in Review)
-
To the best of our knowledge, this is the first empirical study to reveal the current status of deprecated APIs in TensorFlow.
-
Performed the first research discovering the rationale behind deprecated APIs in TensorFlow. We analyzed the deprecation message in 235 deprecated APIs and found 6 API deprecation reasons.
-
Automatically uncover deprecated APIs in 12 existing deep learning models, which helps explore developers’ reactions to TensorFlow deprecated APIs.
-
Present a quantitative study about the impact of deprecated APIs in deep learning models accuracies.
An Empirical Study of the Impact of Architecture Refactoring on Software Performance
- Investigated 46 architecture refactoring related commits from 3 popular Java framework HBase, Cassandra, Hadoop, and classified them into 4 self-defined architecture refactoring categories.
- Run JUnit tests before and after commits to evaluate the performance (CPU Time, Memory Usage, and Response Time) difference.
Xinhua News Future Media Convergence Research Institution Beijing, China 2018.7 - 2019.3
Software Engineer
The Affective Benchmarking of Movies Based on the Physiological Data of Audiences
- Implemented an algorithm which monitors the affective benchmarking of movies based on the physiological responses of a real audience collected from Electro Dermal Activity (EDA) sensor.
- This algorithm could be used to predict movies’ box office, help director understanding audiences’ emotion and improve the plot later.
EDUCATION
Concordia University Montreal, Canada 2019.9 - 2021.9
Master of Science: Computer Science
- GPA: 3.8/4.3
Hunan University Changsha, China 2014 - 2018
Bachelor of Engineering: Computer Science
Coursework
- Implemented a coverage-guided fuzzer with jupyter notebook to test program bugs.
- Design of test case to detect four Java bug patterns defined in FindBugs.
- Implemented one Java project to detect three exception handling anti-patterns.
- Design of sparse matrix-vector multiplication algorithm with MapReduce technology using FLINK.
SKILLS
- C++, Java, Python, R, HTML, MATLAB, MySQL Chinese(native), English