CV
Education
National University of Singapore (NUS), Singapore
- MComp in Artificial Intelligence (1.5 years)
- 08/2025 – 11/2026 (expected)
Beihang University, Beijing, China
- B.Eng. in Software Engineering
- 09/2020 – 06/2024
- GPA: 3.86/4.0, Rank: 4/187
- Honors:
- First Prize of Study Excellence Scholarship (2021, 2022, 2023)
- Outstanding Students (2021, 2022, 2023)
- First Prize, 33rd Feng Ru Cup (2023)
Publications
- Yuansen Liu, Yixuan Tang, Anthony Kum Hoe Tun.
*Reasoning Hijacking: Subverting LLM Classification via Decision-Criteria Injection. ACL 2026. - Yuansen Liu†, Haiming Tang†, Jinlong Peng†, et al.
Human-MME: A Holistic Evaluation Benchmark for Human-Centric Multimodal Large Language Models. ICLR 2026. - Qifan Liang†, Yuansen Liu†, Ruixin Wei†, et al.
Segment-Aware Conditioning for Training-Free Intra-Utterance Emotion and Duration Control in Text-to-Speech. ACL 2026. - Haozhen Gong†, Xiaozhong Ji†, Yuansen Liu†, et al.
Med-CMR: A Fine-Grained Benchmark Integrating Visual Evidence and Clinical Logic for Medical Complex Multimodal Reasoning. CVPR 2026. - Jiangning Zhang, Junwei Zhu, et al., Yuansen Liu, et al.
*Soul: Breathe Life into Digital Human for High-fidelity Long-term Multimodal Animation. CVPR 2026. - Qingwen Yu, Yuansen Liu, et al.
Multi-task learning for calcaneus fracture diagnosis of X-ray images. Biomedical Signal Processing and Control, 2025.
† equal contribution, * corresponding author
Work Experience
TikTok (ByteDance), Machine Learning Engineer Intern
- 02/2026 – 08/2026
- Built and deployed an automated compliance-review pipeline with multimodal LLMs for Proof-of-Address verification.
- Processed large-scale heterogeneous and adversarial real-world image data in a security-sensitive production setting.
- Investigated interpretability/efficiency bottlenecks in black-box models for safer system design.
Shopee, Golang Backend Developer (Full-time)
- 09/2024 – 06/2025
- Designed and launched an automatic promotion renewal system covering 5000+ promotions and 20000+ SKUs.
- Reduced Taiwan operations workload by roughly 50 person-hours/week.
- Redesigned partition strategy and indexes for discount services, reducing request latency to around one-tenth.
CAXA Technology, Research Intern
- 07/2023 – 11/2023
- Constructed a 60,000-scale CAD surface-feature dataset with C++ plugin-based conversion.
- Designed a GNN model for CAD face feature recognition.
- Assisted 3D wireframe fitting to CAD models using pythonOCC.
Skills
- Languages: Mandarin (Native), English (Fluent)
- Web Development: Vue3, Python (Django & Flask), Golang
- Deep Learning: PyTorch, PyG, vLLM
- Game Programming: Java (LibGDX), Unity
- Desktop Development: C# (WPF), Python (PyQt)
- 3D CAD: SolidWorks (Certified)
Standardized Tests
- IELTS: 7.5 (L 8.5 / R 9.0 / W 6.5 / S 6.5)
- GRE: 325 + 3.5 (V 155 / Q 170)
