Nyx Iskandar

🐻 EECS + Public Policy @ UC Berkeley

🤖 AI/ML Research @

+

#about

I'm passionate about studying and improving the fundamental capabilities and intelligence of AI systems to build more reliable and efficient agents, and analyzing and contributing to public policy regarding AI societal risk and security. In pursuit of these interests, I'm actively conducting research in action abstractions for black-box planning agents and data-centric AI governance. I'm also currently building an AI tool for personalized education and contributing to work in assessing the extent to which LLM safety testing must be done. I'm proud to have been awarded the prestigious Smart Nation Scholarship by GovTech Singapore to pursue my undergraduate studies in Electrical Engineering & Computer Sciences (EECS) and Public Policy at UC Berkeley.

Outside of tech, I love to play volleyball, climb/boulder, and draw manga!

#research

Center for Human-Compatible AI (CHAI)

: Studying and designing action abstractions to improve the optimality, reliability, and efficiency of black-box planning agents (advised by Dr Cameron Allen, Stuart Russell Group)

Jun 2024 - Now

Berkeley Artificial Intelligence Research (BAIR)

: Creating, evaluating, and proposing novel technical approaches for data-centric AI governance for the Frontier Data Initiative (advised by PhD student Ritwik Gupta, Trevor Darrell Group)

Sep 2024 - Now

[Independent Research] AI for Education

: Creating an AI pipeline to personalize online textbooks to enhance learner motivation (advised by Prof Gireeja Ranade and Prof Narges Norouzi)

Sep 2024 - Now

[Independent Research] Manga Layout Analysis

: Innovated a novel deep learning computer vision pipeline (instance segmentation and OCR) to automate the semantic and structural analysis of manga

Apr 2021 - Dec 2022

conference paper github