About.
CV ↗I'm Robin—Bin Wang in my publications. I'm an AI researcher and engineer at Apodex in Singapore, working on agent frameworks, multi-agent systems, and pretraining data acquisition and quality.
I received my Ph.D. from USC in 2021, advised by Prof. C.-C. Jay Kuo, after undergraduate study at UESTC. At NUS and A*STAR I²R, I worked on dialogue summarization, multilingual evaluation, and audio-language models. My work combines research, architecture, and implementation, with more than 60 publications.
I grew up in Jining, Shandong, and studied in Chengdu, Hong Kong, Oshawa, and Los Angeles. I moved to Singapore in 2021 and became a permanent resident in 2024. Outside work, I lift weights, read, maintain an audio AI resource hub, and write about career choices and everyday life.
Agent systems & training data
I built a deep-research agent harness from scratch, with multi-source verification, parallel tool use, and multi-agent coordination. My recent work focuses on large-scale pretraining data acquisition and domain- and URL-level quality assessment.
Apodex 1.1 · Apodex-1.0 · FrontierAgent · Apodex 1.1 · mini · Apodex 1.0 · mini
Earlier Agent & Reasoning Research
Earlier work on research agents and reasoning models, including MiroFlow, MiroThinker, and MiroMind-M1, with a focus on tool use, long-horizon reasoning, and evaluation.
MiroFlow · MiroThinker · MiroMind-M1 · Survey: 100 Days After DeepSeek-R1
Audio & Multimodal LLMs
At A*STAR I²R, I led data preparation and evaluation for MERaLiON AudioLLM and co-led model training. I also led AudioBench, covering eight tasks and 26 datasets.
Work
- May 2025 – PresentAI Research Scientist · Apodex, Singapore→
Agent harness architecture, multi-agent systems, and large-scale pretraining data acquisition and quality.
- Apr 2023 – Apr 2025Scientist · Tech Lead (Evaluation & Data), MERaLiON Team · Institute for Infocomm Research (I²R), A*STAR, Singapore→
Tech Lead (Evaluation & Data) for the MERaLiON AudioLLM team under Singapore's National Multimodal LLM Programme, focusing on Southeast Asian audio-language models.
- Sep 2021 – Mar 2023Research Fellow · National University of Singapore (NUS)→
FacEval for factual consistency in dialogue summarization; EvalRank for representation evaluation. Advisor: Prof. Haizhou Li.
- May 2020 – Aug 2020Research Internship · JD AI Research, Mountain View, USA→
Inductive learning for commonsense knowledge graph completion, in collaboration with Stanford (Jure Leskovec group).
- Jul 2016 – Oct 2016Research Internship · Ontario Tech University, Canada→
3D point-cloud hand-gesture recognition with Kinect for human-robot interaction. Mitacs Globalink awardee.
Education
- Aug 2017 – May 2021Ph.D. · University of Southern California (USC)→
Thesis on word, sentence, and knowledge graph representation learning. Advisor: Prof. C.-C. Jay Kuo.
- Sep 2013 – Jun 2017Bachelor · University of Electronic Science and Technology of China (UESTC)→
Undergraduate National Scholarship (2015, 2016); Excellent Graduate of Sichuan Province (2017).
- Sep 2015 – Jan 2016Exchange Student · City University of Hong Kong (CityU HK)→
One-semester exchange program in Hong Kong during undergraduate studies.
Awards
- APSIPA Sadaoki Furui Prize Paper Award (2024)
- APSIPA Sadaoki Furui Prize Paper Award (2022)
- Best Paper Award, SUMEval Workshop at COLING 2025
- Best Paper Award, C3NLP Workshop at ACL 2024
- USC Graduate Student Government Research Travel Grant (2019)
- Excellent Graduate of Sichuan Province (2017)
- Undergraduate National Scholarship, China (2015, 2016)
- Mitacs Globalink Research Internship (2016)
Academic Services
- Publication Chair, EMNLP 2023
- Area Chair, ACL ARR (2024–2025)
- Editorial Board, APSIPA Transactions on Signal and Information Processing, 2023–2025
- Session Chair, IJCNN 2021
- Reviewer: Nature Human Behaviour (2022), IEEE/ACM TASLP, ACL, EMNLP, NAACL, ICASSP, ICME
Students Supervised
Mentored 10+ research students and interns at I²R (A*STAR) and NUS on audio LLMs, instruction tuning, and long-video understanding; many have moved on to strong PhD programs or industry roles.
- Apodex 1.1: Scaling Agentic Intelligence for Complex Work· Technical report, Aug 2026· Contributor, Apodex Team
- Apodex-1.0: A Verification-Centric Agent Team for Discoverative Intelligence· Technical report, Jun 2026· Contributor, Apodex Team
- MiroFlow: Towards High-Performance and Robust Open-Source Agent Framework for General Deep Research Tasks· arXiv 2026
- MiroThinker: Pushing the Performance Boundaries of Open-Source Research Agents via Model, Context, and Interactive Scaling· Technical Report 2025
- MiroMind-M1: An Open-Source Advancement in Mathematical Reasoning via Context-Aware Multi-Stage Policy Optimization· Technical Report 2025
- MERaLiON-AudioLLM: Advancing Speech and Language Understanding for Singapore· ACL 2025, System Demonstrations· Co-first author
- AudioBench: A Universal Benchmark for Audio Large Language Models· NAACL 2025· First author
- CoinMath: Harnessing the Power of Coding Instruction for Math LLMs· Findings of ACL 2025
- Resilience of Large Language Models for Noisy Instructions· Findings of EMNLP 2024
- SeaEval for Multilingual Foundation Models: From Cross-Lingual Alignment to Cultural Reasoning· NAACL 2024· Co-first author
- Knowledge Graph Embedding: An Overview· APSIPA Transactions on Signal and Information Processing, 2024
- An Overview on Language Models: Recent Developments and Outlook· APSIPA Transactions on Signal and Information Processing, 2023
- Compounding Geometric Operations for Knowledge Graph Completion· ACL 2023
- Analyzing and Evaluating Faithfulness in Dialogue Summarization· EMNLP 2022
- Just Rank: Rethinking Evaluation with Word and Sentence Similarities· ACL 2022
- Graph Representation Learning: A Survey· APSIPA Transactions on Signal and Information Processing, 2020
- SBERT-WK: A Sentence Embedding Method by Dissecting BERT-based Word Models· IEEE/ACM TASLP 2020
- Evaluating Word Embedding Models: Methods and Experimental Results· APSIPA Transactions on Signal and Information Processing, 2019
bwang28c@gmail.com