Cheng. Wang
31 | JIT | Chinese | [email protected] | GitHub: Talbot3
Education
Jingling Institute of Technology, Computer Science and Technology Major, bachelor’s degree.
Experience
2022/02 - 2023/03 iSoftStone Information Technology (Group) Co., Ltd. Java Technical Manager
Part of the function design and implementation of the knowledge map team of Shanghai ICBC Science-AI project was transferred to the NLP team of AI project for the development of back-end functions. During Confidentiality training, homework norms training (during the epidemic), Gaussian database training, etc. During my stay at home, I investigated Kerberos\Hadoop, Fink\Spark, Mockito, TensorFlow, Quartz, Jupyterlab, SonarQube and other tools to understand the operator development using H2O-3 in the project and the supporting DAG operator running platform developed here. Thus, after a year of development, the workload was more than double that of the lowest student in the same NLP group. At the handover. When reporting to Party leader, expressed satisfaction and gratitude. Early I am responsible for log data processing optimization and unit test code writing. Late MLOPS epidemic, leading to the beginning of the original AI. The platform is upgraded and communicated with major companies (such as the fourth paradigm company) to understand the design of a new version of the AI middle platform. I am involved in assets of ai (such as datasets, models), data analysis, sharing, process optimization and transformation and so on modules. The experience of using AI in the middle station will be more humanized, and the search and prompt of related functions will be more interpretable. Easy to read and understand, the process is simple and convenient to integrate into the user's operation process. And other written about:
- Write software user manual-PC module (team division of labor to write different module manuals)
- Migrating Deep Learning Training Modules and Related MyBatis Model Configuration (Migrating Spring MVC to Spring Boot Framework)
- Write module test code based on Mockito test framework to ensure that it passes SonarQube code quality inspection platform; at the same time, ensure that the test unit code branch coverage rate is above 95%.
- Due to system evolution, new MLOps and implementations are designed-Due to rapid expansion of the system, log tables need to be cleaned regularly, and configured tasks for cleaning table data templates regularly are designed to improve R & D efficiency- Update vulnerable dependency packages and package detection tools.
- The model built by the migration H2O-3 operator was published on HDFS.It needs to support multiple forms (compressed package, folder), etc.
Later and front-end guys realized the asset search module and transcoded all assets in a unified format.
I shared the design and implementation of visual editing of report technology with my colleagues, which reduced the workload of statistics, which needs to change constantly. duration Several months of design and implementation of the 100+ report statistics interface. This led to more leisure when forced overtime, reading more system code, completing more Mybatis optimization. In log cleaning, a general cleaning command table structure is designed, which greatly reduces the workload of log cleaning. Change writing code to writing SQL Log cleanup table.
2021.10-2021.11 Shanghai Aohe Information Technology Co., Ltd. Intermediate Java Development Engineer
The responsible for connecting the product system of CPIC internal line, analyzing the requirements and guiding the team members to develop.
2019.4-2021.8 DeepBlue Technology (Shanghai) Co., Ltd. Backend Development
Responsible for the design and implementation of the project are: (1) video surveillance analysis system-garbage classification event statistical analysis platform (2) Video surveillance analysis system-luxury residential area behavior event statistical analysis platform (3) IOT card data acquisition system. The above project was engineered using the Pytorch\Nodejs technology stack, during which 1) was noted Tools research, 2)AI Model research and optimization, 3) TCP packet parsing research, 4) Security video streaming protocol GB28181 parsing and rendering implementation and algorithm with the support of students 'collaborative research and leadership, Zhang realized the most from security surveillance camera to GPU server application model for event analysis to visual video analysis system Finally, the leader was invited to an interview. During this period, Talbot3/F2Reactize report was opened on GitHub, rtsp-mock(700+ downloads) was implemented using rust, and video stream decoding and encoding about FFmpeg were shared.
Later project changes, participate in the development of the backend of the education platform system based on Java technology stack. Java technology stack research; research Weixin Official Accounts background technology; research Jenkins, Harbo, Nexus, Keycloak, Nacos, Open feign and other common facilities supporting privatization deployment, optimizing the back-end R & D efficiency Rate; through collaborative development with algorithms, Android, front-end students, etc., from the camera based on Android education laboratory-> Python algorithm analysis system -> Business cycle based on Java education experiment business system.
2017.12-2019.3 Shanghai DaDaABC Education Technology Co., Ltd. Nodejs Engineer
1) Share the enterprise application of K8S and the demonstration of dynamic expansion and contraction. 2) Responsible for the design and development of audio and video client SDK The project is to realize the business by applying audio and video functions in the upper layer of the enterprise. In the early stage of development, the participating teams focused on collaborating to open the SDK of the audio and video client Web version. Later, after stability, Independent with Android\IOS students design and develop SDK supporting H264 codec. Occasionally maintain both WebRtc backend signaling and self-developed SFU. lasted more than a year.
The audio-video-lab team restructured the streaming media service solution, added third-party backup systems and turn servers, and improved the ICE connectivity rate and product availability of each platform (android/iOS/web) of WebRtc client. During my work, I open-sourced Talbot3/getStats analysis tool, investigated FFmpeg\Webrtc\K8s and other tools, cooperated with Android development to analyze Android Audio playback sound interruption problem, shared the knowledge points and use points of Webrtc ICE connection abnormal state in JIRA (Enterprise Knowledge Base), etc.
2017.10 - 2017.11 Shanghai Qingben Supply Chain Management Co., Ltd. Nodejs Development Engineer
Responsible for the preparation of supply chain procurement system, and experienced a distribution process, to understand the operation of the supply chain process. Development Cycle Management Using Agile Management Systems and functional status for daily reporting. Technology stack using Nodejs
2016-07-2017-09 Shanghai equivalent electronic technology co., LTD. Backend Dev Engineer
Responsible for: connecting the hardware of passenger flow statistics developed by students; analyzing the returned data and developing a supporting passenger flow statistics platform according to these data. Divided by Product Manager After analyzing requirements and functions, I design and implement the architecture; the technology stack uses Nodejs.
2015/11 - 2016/01 Naibohu Mobile Technology (Shanghai) Co., Ltd. Software Engineer Intern
Responsible for 66 cell mobile CRM background development: background-API log management; guide page management configuration; H5 maintenance-shielding, top post management; Technology stack is AngularJS and Amaze UI component for front-end web pages in backend management, NodeJS with Async modul in backend, NoSQL with MongoDB as database
Skills
Node technology stack: nodejs, sequelize, nest/react, etc.
Java technology stack: spring-boot, gradle, spring-cloud-alibaba, mybatis, mapstruct, nacos,openfeign, dubbo, maven,Mockito, etc.
Python technology stack: poetry\pipenv\miniconda, flask, sklearn, pytorch, click, numpy, pandas, python-opencv, etc. Database: mysql, MongoDB, Redis, elasticSearch, etcd, etc.
File system: minio (S3) Automation (bash, jenkins, gitlab/github ci, ansible, docker, docker-compose) Gateway: openreplay, nginx, keepalived\ipvs\iptables;etc.
Audio-video technology stack : ffmpeg, mediasoup, coturn, etc.
Familiar with common design patterns and architectural patterns.