Kevin Tirta Wijaya

PhD Student - Max Planck Institute for Informatics

Expert-informed AI in data-scarce domains.

Kevin Tirta Wijaya

About Me

I am a PhD student at Max Planck Institute for Informatics, supervised by Dr. Vahid Babaei. My current research interest explores how domain expertise can be integrated into machine learning systems to improve performance in data-scarce domains like drug discovery and materials science. Previously, I worked on 3D perception for autonomous driving under the supervision of Prof. Seung-Hyun Kong. I earned my M.Sc. in Robotics from KAIST, and my B.Sc. in Electrical Engineering from Bandung Institute of Technology.

Selected Publications

Post Hoc Regression Refinement via Pairwise Rankings

K.T. Wijaya, M. Sun, M. Guo, H-P. Seidel, W. Matusik, V. Babaei. NeurIPS 2025

PDF Code

tl;dr: pairwise comparisons from humans / LLMs can improve a regressor without retraining.

Two‑Stage Pretraining for Molecular Property Prediction in the Wild

K.T. Wijaya, M. Guo, M. Sun, H-P. Seidel, W. Matusik, V. Babaei. ICML GenBio Workshop 2025

PDF Code

tl;dr: a two-stage pretraining improves molecular property prediction performance in very small datasets.

Trustworthy Inverse Molecular Design via Alignment with Molecular Dynamics

K.T. Wijaya, N. Ansari, H-P. Seidel, V. Babaei. Advanced Science 2025

PDF Code

tl;dr: inverse molecular design methods should output solutions with minimal uncertainties to improve trustworthiness.

Differentiable Search Based Halftoning

E. Luci, K.T. Wijaya, V. Babaei. Computer Graphics Forum 2025

PDF Code

tl;dr: a fast, high-quality digital halftoning that optimizes pixels placements in the 3D space.

K‑Radar: 4D Radar Object Detection for Autonomous Driving in Various Weather Conditions

D.H. Paek, S.H. Kong, K.T. Wijaya. NeurIPS 2022

PDF Code

tl;dr: a novel 4D radar dataset for all-weather 3D object detection in autonomous driving.

K‑Lane: Lidar Lane Dataset and Benchmark for Urban Roads and Highways

D.H. Paek, S.H. Kong, K.T. Wijaya. CVPR Workshop 2022

PDF Code

tl;dr: a novel LiDAR dataset for lane detection in low-light conditions.

Contact

I'm always interested in discussing research collaborations, opportunities, or just connecting in general!

Building E1 7, Room 1.15

Saarland Unversity
Saarbrücken, Germany

×

Connect on LinkedIn

LinkedIn QR Code

Scan this QR code with your phone to connect on LinkedIn

Open LinkedIn Profile