- Co-first author, Findings of EMNLP 2026 and LREC-COLING 2024.
- Built 3 open medical NLP datasets: biomedical NER, machine translation and code-switched speech.
- PhD at UTS & VinUniversity on NLP and LLMs for biomedical and low-resource MT.
Nhu Vo
Research Assistant · NLP & LLMs
I build Vietnamese and medical NLP datasets and models, published at EMNLP, LREC-COLING and Interspeech.
Nhu VoResearch Assistant · NLP & LLMsPhD StudentUTS × VinUniversity · Sydney, Australia
Reliable NLP systems, from data to deployment.
I work on NLP for low-resource languages, biomedical NLP and LLM systems. My research focuses on Vietnamese–English machine translation, biomedical NER, code-switched speech recognition, and the data and evaluation practices that make AI more reliable.
- NLP
- LLM
- AI Agents
- Biomedical NLP
- Data-Centric AI
En-ViMedNER · research threadLanguage data
01Language dataEN–VI parallel
02AnnotationUMLS types
03Quality controlflag + review
04Evaluationcross-lingual NER
Aligned sentence pair · illustrative— pending
ENPatients with type 2 diabetes were treated with metformin.
VIBệnh nhân đái tháo đường type 2 được điều trị bằng metformin.
Disease or SyndromePharmacologic Substance
4,392Abstract pairs
44,892Sentence pairs
202,949Aligned entity mentions
Corpus counts · En-ViMedNEREMNLP 2026 Findings
- 5papers at EMNLP, LREC, LREC-COLING and Interspeech (2024–2026)
- 3open medical NLP datasets on Hugging Face
- 8+ yrsuniversity teaching, since 2017
- 4.72/5mean student evaluation (HUFLIT)
- 4competitive scholarships, incl. Vingroup and UTS
Strengths by role
What I bring
- Taught NLP, Python and Health Informatics in English at VinUniversity.
- Lecturer at five universities; designed courses, 4.72/5 student evaluation.
- Supervised 2 theses, guided 65 student research projects, trained 3,000+ teachers.
- Designs LLM agent apps: multi-agent orchestration, hybrid RAG, MCP, human-in-the-loop.
- AI-assisted development with Claude Code, Codex and Orca; specs, tests and review as the gate.
- LLM fine-tuning with SFT, LoRA and DPO; NMT, NER, PyTorch, Hugging Face Transformers.
Selected work
Papers, datasets and systems
Paper + dataset
En-ViMedNER
Findings of EMNLP 2026Co-first author. English–Vietnamese biomedical NER corpus: 202,949 aligned entity mentions, 21 UMLS types, LLM label projection at 98.32% exact match.
Paper + dataset
MedEV
LREC-COLING 2024Co-first author. 358K medical sentence pairs for Vietnamese–English MT; fine-tuned models beat Google Translate in both directions.
Paper + dataset
ViMedCSS
LREC 202634 hours of Vietnamese medical code-switched speech; an ASR benchmark that scores medical terms separately (CS-WER).
Public prototype
Medical speech translation
VinUniversityReal-time speech recognition + medical MT for clinicians and patients. Project Coordinator of the team project; demo built by Nguyen Giang Son.
AI agent app
No More Later
In developmentPriority-planning agent: multi-provider LLM gateway, token budgets, prompt-injection tests, pgvector search, MCP server.
AI agent app
Togo Todo
In developmentShared-planning agent: multi-agent orchestration, hybrid RAG (pgvector + RRF), LangSmith tracing, human-in-the-loop approval.
Interactive papers
Read the research, hands-on
Each paper as a page you can play with: guess, drag and step through its real data.
All publications
PACLIC 34 · 2020Naming the writer of a Vietnamese text from its word habits
13 min read
NICS 2020Counting the words that set Vietnamese stories apart from other texts
12 min read
FAIR 2020Picking 100 million words that stand for Vietnamese
14 min read
Vietnamese Dictionary & Encyclopedia 2020Collecting real Vietnamese text to write dictionary entries from
12 min read
Linguistic Journal 2022Teaching real Vietnamese to foreigners with a 100-million-word corpus
14 min read
HCMUE Journal of Science 2019Writing the script for an online course before it is built
14 min readCollaboration and research
Building NLP or LLM systems for real-world use?
I welcome conversations about AI research, biomedical NLP, dataset engineering, and PhD or research opportunities.