October 2026
The terminal takes the lead.
Explore the current siteI wanted the first screen to show how I think about AI security. The terminal stays at the centre. The surrounding interface makes the simulated attacks easier to discover and explore.
The decisions behind v2
Intent
Make the portfolio feel like an extension of my work in adversarial AI, while keeping the projects, research evidence and existing interactions accessible.
Keep the terminal focused.
Scenario choices and input explanations live outside the terminal. Inside it, the bash-style output has one job: show the simulated execution.
Make discovery automatic.
The scenarios loop by default. Visitors can choose an attack, inspect a stage or pause the playback. A manual selection does not end the loop.
Let motion support the content.
I removed the rotating circle because it competed with the terminal. A quieter background responds to the pointer and scroll, with a static treatment for reduced motion.
Design the whole experience.
The chatbot, research evidence, navigation and dialogs use the same visual language. Light mode uses a milky cream base. Both themes let the background continue between sections.
Tradeoff
More animation is not always more useful. I kept the interactions that help visitors explore the work and removed decoration that weakened the hierarchy. The terminal remains a simulation, clearly labelled as such.
Evidence & limits
The review includes light and dark themes, desktop and narrow mobile layouts, terminal playback, research filters and evidence images, chatbot interactions, navigation and dialogs. Automated checks cover the build and the existing test suite. These are local checks, not a claim of measured visitor impact.
Screenshots capture this design iteration. Portfolio content continues to evolve.



