Autonomous Driving - Middleware

Autonomous Driving - Performance Testing Platform

Autonomous Driving - Tracing Infrastructure

Backend Systems @Metropia Building Data and Machine Learning Pipeline + RESTful API with scalability, availability, reliability and low latency in mind; Test and document driven development; Deployment and operation on AWS;Stack: Python3, Django, Django Rest Framework, PostgreSQL, Redis, Celery, Pytest, JupyterLab, Scikit-learn and Pandas.

ANTARES Project (https://www.noao.edu/ANTARES/) Main developer and maintainer of ANTARES (The Arizona-NOAO Temporal Analysis and Response to Events System) project. Deployed ANTARES system to the dedicated CentOS cluster using MySQL Cluster as Database backend. Analyzed and improved the performance of ANTARES data processing pipeline. Researched data provenance. Designed and proposed the data provenance framework for ANTARES to answer provenance questions that astronomers would ask. Designed the packet format specification for alert data and implemented the simulator for generating alert streams in Python.

Named Data Networking (https://named-data.net/) Project Researched consumer-driven congestion control mechanisms in NDN. Implemented TCP-like congestion control algorithms (RENO, CUBIC and VEGAS) in NDN consumers (in C++) to lay the groundwork for comparison with more advanced congestion control algorithms for NDN.

TA for CS460: Database Systems Held office hours; graded programming and written assignments; prepared solutions for written assignments. Designed the final project "Database-driven Web Application" using Oracle, Tomcat and JSP; evaluated students’ design and implementation.

ANTARES project (https://www.noao.edu/ANTARES/) Designed and implemented the database used by ANTARES, as well as the web front-end for querying data produced by ANTARES pipeline using Django and MySQL. Collaborated with astronomers to design, implement and document the Python API for astronomical alert data manipulation. Designed and built the auto-configuration and bootstrap system to run ANTARES in a pseudo-distributed mode using Puppet and Vagrant.

Large-scale Neural Modeling in MapReduce and Giraph (http://imsure.github.io/NeuralGiraph/) Researched graph processing in Hadoop; proposed an improved graph algorithm design pattern in MapReduce. Implemented a large scale basal ganglia neural network model in MapReduce. Implemented the model with Apache Giraph, improved performance by 60% compared to the MapReduce implementation.

 Migrated the legacy software to the new hardware platform with other team members and conducted unit testing. Improved overall software quality using static code analyzer FlexeLint, identified and corrected several vulnerabilities existed in the legacy code. Built a tool (in Python) for summarizing, indexing and querying large volume of warning messages produced by FlexeLint.

 Developed Linux device drivers and applications (in C) for various ARM platforms (ARM 9, 11 & Cortex-A8). Assisted training instructors in preparing training content and developed technical how-to documents for ARM and Embedded Linux training. Instructed trainees in their lab sessions and final design project.

 Being part of a team that built the initial prototype for Samsung’s Remote Test Lab (RTL), I was responsible for the development on RTL device. Implemented the communication protocol (in C++) between RTL device ( LiMo platform) and RTL proxy. Developed a daemon process (multi-threaded) in C++ for RTL device to handles requests from proxy, in order for RTL client to remotely control the device.