Vissuta Gunawan Lim
CS Undergraduate
CS Undergraduate at the University of Indonesia with research interests in multilingual agentic evaluation and automated radiology report generation. Contributing to agentic LLM benchmarking for Southeast Asian languages at SEACrowd and building CT-based report generation pipelines at MLCV Lab. Hands-on experience in LLM evaluation, multi-agent systems, and medical computer vision, collaborating with international research teams across Southeast Asia.
News
🎉 SEATauBench accepted to EMNLP 2026 Findings! arXiv
Joined SEACrowd as a Research Apprentice, mentored by Samuel Cahyawijaya and Patomporn Payoungkhamdee, working on multilingual agentic evaluation for Southeast Asian languages. My first research to date, looking forward to what lies ahead!
Research Interests
Multilingual Agentic Evaluation
Benchmarking agentic LLM capabilities across low-resource Southeast Asian languages, focusing on tool-use evaluation and multilingual adaptation, as part of the SEACrowd international research collaboration.
Mechanistic Interpretability
Opening up model internals rather than scoring their outputs: linear probing, attention analysis, and activation steering on a volumetric CT report generator, and turn-level behavioural analysis of what small agents actually do as they scale.
Radiology Report Generation
Automated CT scan report synthesis for lung nodule detection at MLCV Lab, evaluating encoder-decoder and VLM-based architectures on clinical accuracy and latency trade-offs.