(co-corresponding)
ranked-list answer format evaluation (e.g. MedQA) for medical reasoning models in collaboration with Typhoon.
tossaporn.sae [at] mahidol.edu
data scientist @
SiData+,
Siriraj Hospital
MIT EECS 2019, MEng 2020
cv
linkedin
github
machine learning, healthcare, llm, vision, generative ai.
VeReaFine: Iterative Verification Reasoning
Refinement RAG @ACL BioNLP 2025
iterative Verifier-RAG pipeline for open-ended clinical QA.
Decomposing
Food Images for Better
Nutrition Analysis @CVPR MetaFood 2025
(corresponding)
two-step multimodal LLM on Nutrition5k and Gindee121.
Data Scientist at SiData+, Siriraj Hospital
(co-first)
robust score distillation technique for image editing and
object insertion while preserving image backgrounds
[project page]
GPU Poor’s Guide to Radiology Report Generation
@ACL BioNLP 2024
lightweight multimodal language model with a two-stage post-processing for radiology report generation
Research Assistant at Vision and Learning Lab, VISTEC
coached Mok Wattanasopon to compete in the IOI 2022
taught computer science, mentoring award-winning student AI projects
Graduated, Computer Science Teacher, MWIT (national science highschool)
built stuff on crypto payments, music sharing, and network visualization
internship at Facebook (now Meta) in 2017
Admitted to MIT
bronze medal in the IOI 2013