About
I am a Computer Science Grad Student at University of Southern California. I enjoy problem-solving and coding. Always strive to bring 100% to the work I do. I have worked on technologies like Python, Java, C++, Django, Flask, MySQL, PostgreSQL, MongoDB, HTML5, CSS and so on. I have 12 months of professional work experience which helped me strengthen my experience in software development.
I am passionate about software engineering and open source, and I am always eager to learn new technologies and apply them to real-world problems. I am looking for opportunities to collaborate with other engineers and developers, and to contribute to the advancement of the software industry.
I am graduating in 2023 December, currently looking for an entry level software developer position.
Experience
- 222 is a Y Combinator backed social experience startup. I worked on the backend team to build a scalable and robust backend for the 222 app.
- Led the development of a robust backend for venue scraping with Python's FastAPI, integrating Yelp and Google Maps APIs, which facilitated targeting high-traffic venues, leading to a 25% increase in footfall.
- Designed a scalable MongoDB database architecture that reduced memory usage by 50% and increased query speed.
- Used OpenAI's GPT-3.5 API to generate venue descriptions from Yelp and OpenTable data, enhancing user experience on landing page.
- Tools: Python, FastAPI, Flask, MongoDB, Openai API, AWS, Docker, Selenium, BeautifulSoup, Git, JIRA
- Designed and executed a Bash solution to integrate Kyverno policies onto Artifact Hub and reduced manual intervention by 70%.
- Devised CI/CD system and rewrote Github Actions workflow to ensure new policies can be easily added into Artifact Hub in future.
- Analyzed user feedback and updated documentation to improve clarity and effectiveness, reduced support requests by 20%.
- Tools: Shell, Python, CI/CD, Github Actions, Git, Kyverno, Artifact Hub
- Initiated a domain adaptive object detection project and updated models from state-of-the-art models using PyTorch and Detectron2.
- Increased mean average precision for rare classes by 5.9 points through combination of self-training task and multi-level feature alignment.
- Reduced over 50% training time for large datasets by means of distributed data parallel library.
- Tools: Python, Deep Learning, PyTorch, Detectron2, Docker, Git
- Orchestrated full-stack development of a travel booking website using Node.js, Express.js, and React.js for interactive user experience.
- Leveraged AWS services (EC2, S3, RDS) for application hosting, data storage, and user authentication.
- Utilized Hadoop and MapReduce on AWS EMR for parallel execution of K-means clustering algorithm, improving model accuracy by 8% and optimizing resource utilization.
- Tools: Node.js, Express.js, React.js, MongoDB, Hadoop, MapReduce, AWS, Git
Projects
COVID-19 tracker and analysis
- Tools: Angular, Node.js, Express, Mocha, Chai, Bootstrap4, TypeScript
- Engineered a real-time virus data visualization tool with Angular and Bootstrap4 and updated automatically every hour from server.
- Developed and deployed a Node.js and Express server that enabled real-time communication between frontend interface and server.
- Conducted 50+ unit tests using Mocha and Chai to ensure efficient code execution and reduced debugging time.
Mini Stackoverflow Website
- Tools: Vue.js, Heroku, HTML, CSS, JavaScript, Bootstrap
- Created website pages showing the basic function of Q\&A website using Vue.js
- Conducted user authentication, authorization and backend for posting and answering questions
- Performed rapid prototyping and testing with team members while demonstrating iterative design strategies
- Reviewed code for browser compliance and fixed bugs. Hosted website on Heroku.
Naive Docker Implemented by C++
- Tools: C++, Docker, Linux native API, namespaces, control groups, Kubernetes
- Developed a Naive Docker in C++, leveraging namespaces and control groups for resource isolation and containerization.
- Utilized Linux native API to manage network operations within the containers and effectively control container resources.
- Contributed to the development of a test environment using Docker containers and utilized Kubernetes for container orchestration.
improve the algorithm that classifies drugs based on their biological activity
- Tools: Python, Deep Learning, Transfer Learning, Data Analysis, Ensemble Model
- Explored data analysis and feature engineering with Rank-Gauss, PCA and variance threshold.
- Performed transfer learning with three different CNN model and used blending as model ensemble method.
- Developed algorithm to predict a compound's MoA given its cellular signature, thus helping advance drug discovery process.
Fast, Realible File Transfer with Custom TCP/IP
- Modified TCP module of networking stack of Linux kernel (3.2.0) to optimize file transfer.
- Removed exponential back-off and slow start phase; Achieved 4Mbps throughput on a 100Mbps link with 20% loss and 400ms RTT.
- Set up static routing on a small network of 4 hosts and 3 routers and properly exchange packets on AWS.
Computer Vision Project: Crowd Density Estimation
- Tools: Matlab, Caffe, LMDB, Hdf5
- The Crowd-Density-Estimation project aimed to estimate crowd density from images using a deep learning framework, Caffe. Instead of using the shanghaitech dataset mentioned in the original paper, the malldataset was used, and the labels were entire images instead of 0 and 1, which was a unique approach encountered for the first time.
- The project involved writing a MATLAB program to generate MCNN density maps, preparing the data, and then training the model using Caffe. Initially, hdf5 was considered for data transmission, but due to its slow speed, two lmdb were used, one for data and another for labels.
- During the testing phase, mean subtraction and normalization were performed to speed up model convergence. The mean was set to 127.5, and normalization was done by dividing by 128. The final test program was written in MATLAB, and the Caffe framework was used to load the trained model and test it on a new image.
Fullstack Business Search and Reservation Application (Android and Web)
- Architected a business search and reservation application for both Android and web, adhering to MVVM Architecture Pattern.
- Leveraged Yelp API for autocomplete input suggestions and integrated map locations and reviews via Android's Tabbed Activity.
- Ensured user data persistence through Android's Shared Preferences and created a responsive, mobile-first web front-end with Angular.
- Hosted server-side applications on GCP, effectively tested for reliability and functionality, and facilitated social sharing.
Skills
Programming Languages
Libraries
Frameworks
Other
Education
University of Southern California
Los Angeles, USA
Degree: Master of Science in Computer Science
GPA: 3.5/4.0
- Distributed Systems
- Internet and Cloud Computing
- Analysis of Algorithms
- Parallel and Distributed Computation
- Web Technologies
- Computer Systems Organization
- Database Systems
- Information Retrieval and Web Search Engines
Selected Courseworks:
Wuxi, China
Degree: Bachelor of Engineering in Computer Science and Technology
GPA: 3.5/4.0
- Machine Learning
- Compiler Systems
- Operating Systems
- Data Structures
- Database Management Systems
- Computer Networks
- Introduction to Algorithms
- Computer Vision
Selected Courseworks: