Build for the operating reality
Useful software has users, data, constraints, and failure modes. I care about all four.
About
My Journey.
How it started
One Dell desktop, Windows XP, a PTCL EVO modem, and Project IGI.
Before Kubernetes, there was a Dell PC and a lot of curiosity.
My earliest memory of a computer is a Dell desktop my uncle brought home when I was six. It ran Windows XP, and my older brother and I shared it constantly. I did not know a thing about its specifications. I only knew that pressing its power button opened a whole new world.
We spent hours starting it up, watching films from CDs, and seeing my uncle reinstall Windows from discs. A PTCL EVO USB modem gave us a fragile but exciting connection to the internet. We learned to connect it, browse, download games, and play Project IGI.
By sixth grade, that curiosity had become a direction. Computer science was the field I wanted to understand deeply, and that thread followed me through school, matriculation, and ICS. It made computers feel less like appliances and more like systems worth taking apart, tracing, and improving.
In college, I installed Turbo C++ on my Windows 7 PC and began writing and running C programs. Seeing a small program compile and do exactly what I had asked it to do was exciting. It turned curiosity about computers into a habit of building with them.
That moment did not choose a path for me all at once, but it gave me a lasting habit: trace a system, understand how it works, and try to make it better.
How the work evolved
From business software and AI reasoning to cloud-native infrastructure.
Each step moved me closer to the systems underneath.
That habit became professional at Maima Soft, where I started as an intern and became an associate software engineer in under a year. I worked on enterprise accounting, pharmacy, and school-management products. These systems need to stay dependable in the middle of everyday operations.
At Turing, I moved closer to AI reasoning work, designing mathematical and algorithmic challenges alongside production-grade C/C++ reference solutions. It sharpened the way I think about correctness, constraints, and the difference between a plausible answer and a technically sound one.
I then took that systems mindset into the cloud-native ecosystem. Through upstream contributions to Volcano and Kthena, I have worked on scheduling correctness, resource accounting, controller reliability, and infrastructure for AI inference workloads on Kubernetes.
Today, as an LFX mentee with CNCF Volcano, I am designing topology-aware placement for accelerator workloads. Alongside this work, I mentor students and junior developers at Dev Weekends, sharing the same problem-solving and full-stack foundations that shaped my own start.
Contribute where the foundations matter
I learn by tracing real systems and then turn that understanding into upstream contributions.
Explain the machinery
Writing is how I make a system legible enough to improve, share, and operate.