Vector databases are the infrastructure of the AI era. From Retrieval-Augmented Generation in LLMs, to multimodal recommendation, to anomaly detection in finance — every modern AI system relies on fast, accurate similarity search over high-dimensional embeddings. This ANU-based PhD position invites you to build the next generation of vector database systems.
Research Direction
Embedding unstructured data (text, images, audio) into high-dimensional vectors has become the default representation for AI workloads. The system that stores, indexes, and serves these vectors is the vector database — and there are many open problems. The student will pick one or two of the following directions based on their interests:
- Streaming and dynamic indexes. Real-world vector datasets are continually inserted, deleted, and updated. How do we maintain high-recall ANN search without expensive offline rebuilds?
- Hybrid queries. Most production queries combine vector similarity with structured filters (location, time, attributes). How do we co-design index structures that handle both efficiently?
- Disk-aware, billion-scale indexes. When the dataset exceeds memory capacity, the index must live on disk while still serving millisecond queries. What is the right cost model and data layout?
- GPU-accelerated ANN. How do we exploit modern hardware (GPUs, NPUs, specialized accelerators) for both index construction and query processing?
Supervisor and Benefits
Primary supervisor: Dr.Mengxuan Zhang (ANU School of Computing) — vector database, ANN search, high-performance query processing
Co-supervisors may be added depending on the chosen sub-direction.
ANU PhD stipend: AUD39,069 per annum (Full-time base stipend rate 2026), and travel support for top-tier conferences (subject to research outcomes)
Eligibility
- Bachelor’s or Master’s degree, ideally with at least Second-Class (Upper) Honours or equivalent
- Strong programming background — C++ strongly preferred
- Solid foundation in data structures and algorithms
- Familiarity with machine learning / deep learning fundamentals
- Genuine interest in systems research and a willingness to engage with low-level engineering when needed
- Open to both international and domestic applicants
What you will get from this PhD position
- Weekly one-on-one meeting with hands-on research guidance from Dr. Mengxuan (you will also have another two supervisors in the supervision panel)
- Great change to have top-tier publications(VLDB, SIGMOD, ICDE), if you are self-motivated and dedicated to research
- International research network around Dr. Mengxuan
- A pathway to either academic or industry careers in AI infrastructure
How to Apply
Please email Dr. Mengxuan Zhang at Mengxuan.Zhang@anu.edu.au with the Subject line (which is “PhD Application in Vector Database (ANU-based)”), your CV, Academic transcripts(undergraduate and any postgraduate), a statement (one page or less) describing your research interests, and names and contact details of two academic referees.
Strong candidates will be invited for a video interview. Applications are reviewed on a rolling basis — we encourage early submission.