CV
You can also download my CV here!
Xingyue Huang
Tel: (44) 7579 902135
Email: xingyue.huang@cs.ox.ac.uk
Website: https://hxyscotthuang.github.io/
Education
- University of Oxford
DPhil in Computer Science (2023 – 2026)
Advisors: Prof. Michael Bronstein (DeepMind Chair of AI), Prof. Ismail Ceylan - University of Oxford
MMathCompSci in Mathematics and Computer Science (2019 – 2023)
Graduated with Distinction
Professional Experience
- Snap Inc.
Research Intern (UMaP) (06/2025 – 10/2025)- Developed Threshold Differential Attention (TDA), a sink-free, ultra-sparse attention mechanism for long-context LLMs, achieving >99% exact zeros while matching Softmax accuracy on QA benchmarks.
- Demonstrated long-context robustness on the SCROLLS benchmark and passkey-retrieval stress tests, where TDA outperformed Softmax by ~2.5× at 4k-token contexts.
- Co-authored Hierarchical Token Prepending, improving long-document embeddings in decoder-only LLMs via block-level summary tokens, with 5% gains across 11 retrieval datasets and 30 embedding benchmarks.
- Eigent-AI
Research Intern (10/2024 – 06/2025)- Built a tool-use synthetic data generation pipeline producing 20k verified execution traces for CAMEL-AI.
- Used back-translated tool trajectories for supervised fine-tuning, improving math benchmark accuracy by 5%.
- Led the Loong verifier-driven RL framework for long chain-of-thought synthesis.
- Alibaba Group
Machine Learning Engineer Intern (07/2021 – 09/2021)- Developed an object detection system for video subtitle-detection with Faster-RCNN model.
- Improved accuracy of object detection and classification by 10% and were incorporated into production.
Technical Skills
- Systems & Infrastructure: Triton, LLM architecture design, multi-agent systems, post-training
- ML & Data: PyTorch, HuggingFace, CAMEL-AI, SQL, DuckDB, pandas, torch-geometric
Teaching
Services
- Reviewer: NeurIPS 2025 Top Reviewer · ICML 2026 Silver Reviewer · ICLR 2026
- Workshop Organizer: NeurIPS 2025 SEA, ICML 2026 GFM
- Talks: Tutorial on Graph Foundation Models (LoG 2025) · Relational Hypergraphs (TU Wien Guest Lecture on GRL 2026/03/19, LoG 2024 Oxford Meetup)