Jiaxin(Jason) Li

Microsoft

Azure Chaos Studio (Aug. 2022 – Present)

• Azure Chaos Studio helps to improve application resilience with controlled fault injection and chaos testing by introducing apps to real or simulated faults.
• Took part in the development of web API and backend system for faults life-cycle management (create, start, stop, dispose) and telemetry monitoring system.
• Implemented several machine level faults (memory/CPU/disk pressure, time change) in both Linux and Windows Server systems using C++/C#.

The Michaels Companies, Inc.

Backend API Development – Michaels.com Ecommerce (Feb. 2021 – Aug. 2022)

• Participated in the design, development, improvement, test, and deployment of the high scale backend services for the entire online shopping experience.

Third-party Service Integration (Nov. 2021 – Aug. 2022)

• Participated the integration of the third-party service including Affirm (financing), Attentive (text message), Fusion 92 (promotion), Tibco (loyalty), and SheerID (ID validation) etc.
• Took part in the implementation of the global third-party service admin portal. Helped to create Java Spring tool package for third-party API metric/usage/access/defect/data analysis.
• Led the development of the data retrieving APIs and data uploading/syncing up pipelines for 3rd-party service data access.
• Introduced cache system to store the frequently accessed data and avoid the entire list traverse. Greatly reduced the average API data retrieving time from 1000ms to 10ms.

Shopping Cart Development (Feb. 2021 – Oct. 2021)

• Designed, developed, tested, and deployed of the shopping cart system and RESTful APIs using Java, Spring Boot including initialize shopping cart, add/remove product, change quantity, inventory check, etc.
• Worked with the front-end team to discuss API design such as HTTP method, request/response format, and error handling, etc. Under the premise of realizing functionalities, optimized the structure of request and response (params/JSON), reduced the amount of transmitted data and achieved faster API access.
• Incorporated Kafka as message broker and data buffer for decoupling checkout and shopping cart microservices. Implemented shopping cart cache with Redis which increases performance and reduce response time by 92%.
• Cooperated with the DB team to migrate item table from MySQL to MongoDB using primary-secondary mechanism. Constructed the new shopping cart table schema and index strategy. Managed to merge the separated shopping cart tables (course products, project products, and SKU products) into one standalone table by normalizing the optional tables and extracting the same entities. Reduced the average API long queries from 3 times to 1.5.
• In charge of: using (TDD) Junit and Mockito to keep unit tests light and non-invasive, using SonarLint for code style check, writing swagger API doc, conducting the CI/CD pipeline from dev to QA, and integrating Datadog for API monitoring.

iOS Mobile Development (Dec. 2020 – Feb. 2021)

• Worked on a 16-people mobile development team to design, develop, test and deliver Michaels apps in an agile development cycle. Participated in the start-up of iOS team. Collaborated with team members to research different mobile development frameworks (hibernate vs native, storyboard vs SwiftUI), iOS programming languages (Swift, OC), decide coding style, code structure pattern, version-control flow, third party libraries, and CI/CD pipeline.
• Contributed critical, shared components to multiple features of the mobile app including reusable UI component library, global data transformation, presentation, cache, persistence, and network communication library, etc.
• Closely worked with backend team to discuss API design and requirement, and UI/UX team to design and integrate the app’s UI and prototype using Sketch and Figma.
• Responsible for the development of instore scan & go app from scratch. Implemented the use of WIFI SSID, Bluetooth beacon, and Geofence to improve the user's instore status detection speed. Researched common barcode formats, and compared Apple's official Vision barcode scan solution with Google's MLKit solution to improve scanning and product information retrieving speed.

Environmental Water Quality Monitoring

A real-time water quality monitoring system by image recognition

Code Repo Docker Container

• Researched common methods (Laboratory-based, Test-strip-based, Deep-learning-based etc.) to provide a rough estimation of water sample quality based on water image.
• Helped the research team to develop a water quality estimation API and an iOS app for user to upload water image and get estimation value.
• Created Python Flask RESTful API service based on Google Cloud Compute Engine integrated with water deep learning model (Keras) to supporting water image recognition and HTTP requests/responses.
• Integrated the application with Google Firebase SDK which provides data persistence (NoSQL Database) and user authentication.
• Deployed the web API with Gunicorn, Nginx and packaged as a Docker container.
• Reduced image detection time by 90% by converting the deep learning model to CoreML that can be used directly on the local device and avoid the time-consuming upload task and inevitable network interrupt error.
• Developed an iOS app to implement taking, editing, managing, uploading, persisting photos, and presenting data functions.
• Communicated with UI team to design and integrate the app’s user interface and prototype using Sketch and Photoshop.
• Improved the app’s performance by concurrency and multi-hierarchy cache using Texture and SDWebImage.

Lite OS File System

Hands-on exFAT Linux-like File System

Code Repo

• Closely modeled the original Unix Filesystem and the Unix Fast Filesystem.
• Designed the file system layout and data structure of inode, bitmap, block store, primary/secondary/... directory, and file descrpitor.
• Implemeted the file sytem formatting, mounting, and unmounting functions as a block storage.
• Implemented the creation of files and directories, the opening and closing of files, and the enumeration of directorie.
• Implemented the reading and writing of data to a file, the seeking of file descriptors, and the deletion of files as well as empty directories.
• Implemented the moving files and directories as well as the creation of hardlinks.
• For the file system layout, please see the document.

pokemon Classification

Mobile App for Pokemon Classification

APP Preview

• Created a Convolutional Neural Network using Keras & TF to recognize Pokemons.
• Used Bing Image Search API to generate the Pokemon datadest.
• Converted the Keras model to CoreML model and integrated to mobile app.

Mizzou Scheduler

Calender-like App helps people to schedule ther daily activities

APP Preview

• Helped Mizzou student get their class schedule, grade, GPA with one click instead of complex operation at Mizzou official website.
• Integrated Myzou system API to automatically import student courses, GPA and other information.
• Designed algorithms to help students automatically arrange their courses, assignments, custom projects and other activities.