Research group 002 / AI systems and open source
HKUDS
A research-fit simulator for a prospective PhD conversation around LLMs, AI agents, graph/RAG systems and public open-source engineering. PhD 001 provides a compact research-fit lab; Package 007 expands that into proposal, benchmark, feasibility and supervisor-panel defenses. Both use fictional cases rather than private interview material.
Active simulation
PhD 001 / Research fit
Four original pressure rounds / designed, unmeasured
From public work to a falsifiable PhD plan
Practice peer evidence review, graph/RAG system reasoning, agent evaluation, and a PI plus academic-leadership research-plan defense. The roles are fictional training lenses; they do not impersonate any HKUDS member, Professor Chao Huang, or HKU leader.
- Open-source evidence
- Graph/RAG
- Agents
- Evaluation
- Research plan
Proposal to supervisor-level defense
Defend research motivation, a temporal graph-memory hypothesis, controlled baselines and ablations, negative results, feasibility, ethics, and the current application boundary. No publication, supervisor endorsement, funding or admission is implied.
- Fit
- Temporal graph
- Ablation
- Evaluation
- Ethics
Public-context boundary
Research themes, not borrowed authority.
Public sources identify Professor Chao Huang's work in LLMs, AI agents and graph machine learning, and HKU CDS's cross-disciplinary computing structure. PIT uses that context to design transferable preparation; it does not predict questions, criteria or outcomes.
Public research context →