← Blog

I submitted DRIFT to DSN 2026

My first work as an undergrad researcher

Today I submitted DRIFT to DSN 2026 — the first paper I've worked on as an undergrad researcher. It was an honor just to be able to submit to a conference like this.

DRIFT is about detecting domain generation algorithms (DGAs). The hard part isn't catching today's malicious domains — it's that the algorithms keep evolving, so a detector that works now slowly degrades over time. That problem, concept drift, is exactly what the model is built to survive.

DRIFT has two backbones, each needing its own pre-training and fine-tuning, so it took an enormous number of experiments. Near the end I was driving five machines at once, keeping every run going in parallel.

I gave it everything for days on end, and finished by staying up through the final night. I'm proud that I got it done.

And I owe it to two people: my dear colleague Chaeri, who brought this kind of passion out of me, and my advisor Prof. Jeong, who always has guided us in the right direction.

Hope it goes well.

More from these days

Outfit check on the way to the lab.Another late night — the three of us heading home together, already the next day.Coding even on the move — laptop out in the elevator. (3 a.m.)The screen after a run ballooned in memory and hit an OOM.Midnight on the stairs, waiting for a late-night snack delivery.Chaeri’s notes, not mine — I photographed it because we were in the same boat.Runs and plots, a few days out. Controlling 5 computers & GPUs.Pointing through the paper on submission night.The lab on the last evening, before we hit submit.Email confirming the submission: Submitted #392.The lab, late on submission night.

hover to play

|