I am currently working as an ASML HPC engineer writing automation scripts in San Jose.
Gathering tools to push Artificial Intelligence to be General.
I Love automation - Productivity Increase Always Wins.
Artificial Intelligence
Tensorflow, PyTorch, Sklearn
HPC
Lustre, Ansible, self-developed CAS package ( including multiSSH3 ), Prometheus
OS
RHEL 7/8/9 ( and fedora), Debian (Ubuntu) (Proxmox), Arch Linux, Windows
Coding
Python3, Bash, JavaScript, Java, C
Container
Docker, KVM + QEMU
Packaging
pip, yum, docker, nuget
Web Development
FastAPI, Flask, Fjango, jQuery, Angular, express + axios, Tailwind, Plain JavaScript + HTML
Language
Fluent in English and Chinese, Able to speak Japanese
An complete replacement of Anisble. At least 10x faster ( sometimes more than 1000x faster , use less resources, use more general libraries)
On par or faster than PDSH
Able to produce live outputs with color ( using curses )
Able to aggregate outputs
Able to output json / greppable / summarized outputs
Able to live input characters ( in curses session ), also cache input for waiting hosts
Able to handle more than 1000 parallel hosts
Very highly paralleled: able to use > 8000% of CPU to establish connections / transfer files
~ Infinity size waiting pool
Able use use 2 level resolve for hostnames and hostname groups
Single importable script. No other dependency.
Able to generate / store default values in config files
...
Self-tuning highly parallel copy designed for RAID and SSD systems.
Many many features related to moving file from one place to another in parallel.
Loop device + DD + CP, highly parallel and more easy to use
multiCMD - A simple script that is able to issue multiple commands and execute them at the same time locally.
simple-iotest - High performance minimal io benchmark
TeeLogger - Log and also print to screen with color
TSVZ - TSV Zed - Read / Write TSV, Also able to use one as a in memory delayed sync key-value database.
statblk - Gather disk and partition info for block devices.
statbtrfs - Check Btrfs filesystem status. Able to issue scrub commands in parallel and monitor scrub status while processing.
Currently employed as HPC engineer.
Developed CAS Cluster Automation System to automate cluster operations minimizing downtime and human intervention. Built automation script lib to support cluster automation scripts.
CAS system is designed to enable all personal (including less technicly abled ones) to complete following tasks quickly and effortlessly
monitor
deploy
update
manage
change config
self-heal
start / stop
benchmark
test
access
backup
image ( clone )
live restore
Many more stuff : ) Please ask me for details if you are interested.
Completed Master of Science in Computer Science (Artificial Intelligence) with high GPA in 1.5 year while spending 0.5 year at ASML
Completed half year internship with hands on experience with company projects involving Legal Intercepting Software.
Done scripts simulating, generating, capturing RADIUS traffic, simulating router CLI, server management and deployment in RHEL.
Being student instructional assist for CSE 2431 System II: Operating Systems for two semesters across 4 sections.
Completed various practices during course works at OSU with five dean’s list qualifications and an overall/major GPA of 3.72/3.84
Participating in the Cincinnati Zoo’s project designing and building automatic RFID involved feeding scheduling and information recording system. In particular, designed, developed and deployed the physical system hosting on many Raspberry Pi 4s interpreting instructions and data from multiple RFID antenna and backend server, executing tasks to multiple feeders, recording to internal database while automatically maintaining system autonomy and integrity even if server is disconnected or file is corrupted. Also responsible for creating a docker container for easy deployment hosting a Django server and database created by teammates.
Participating in a project at OSU creating a containerized IOT honeypot and gathering hacker data via hidden loggers. Responsible for creating loggers gathering data from the docker container and KVM and storing them to a hidden secure database. Also responsible for renting and setting up the Kernel Virtual Machine we are using as well as the docker environment for the simulated IOT device.
Practice hardening and defending operating systems at OSU VMWare environment from other students during information security project course.
Created a script for easily retrain a GPT-2 model using provided text library for generating text from prompt during speech & language processing.
Software developer intern for a summer
Developed a webpage-based client doing only Peer to Peer instant messaging app for internal communications within the company.
Software Research intern for a summer
Participated in a project experimented with the Directed Energy Deposition (DED) method for rapid metal prototyping (commonly known as Metal 3D printing). It uses sustained high-power laser beam and metal powder jet to achieve rapid manufacturing of high strength alloy metal prototypes).
Helped to develop an online CCD-observing system that monitors pool status at the depositing site.
• RADIUS Traffic Simulator – At SS8 Networks Inc.
It consists a server and multiple clients running freeRADIUS. They will use scripts I wrote to simulate legit RADIUS traffic complete with client ID and mock traffic data. Traffic is then captured using wireshark and replayed on our test network to test our product before handing over to customer. Project have information specific to customer and stored in company database thus no links here.
Skills involved: Python; WireShark; FreeRADIUS
• NOKIA VSIM CLI INTERFACE – AT SS8 NETWORKS INC.
This is a CLI interface completely written in python in order to simulate the NOKIA VSIM interface for their high performance routers. Our company product will interact (SSH into) with such interface and configure parameters and targets in it. I was tasked to create such a simulated interface so we can test functionality of our product. Interface completely written in python and will give desired output when given input. Able to function standalone and running in Docker container and able to function as a interface over SSH.
Also able to store configured data in csv for verification. Program uses stacked folder structure and multiple standalone scripts to simulate bizarre behavior of said interface. Project have customer information and stored on company server thus no links here
Skills involved: Python
• HUNGARY ELEPHANT – CAPSTONE COURSE AT OSU
As described above in experience section, an automated complex trigge feeder system was developed
Skills involved: Python; Django; Raspberry Pi; systemd
https://github.com/yufei-pan/Hungry-Elephants
• IOT HONEYPOT – NECTWORK SECURITY COURSE AT OSU
As described above, a project focused to gather data about hackers we lured with an honeypot on my rented VPS.