CV

LLM Benchmark Auditing — Surface-Feature Leakage (TruthfulQA-476)

• Discovered that a top LLM truthfulness benchmark (TruthfulQA) is gameable by answer style alone: a question-blind six-feature model (SURFACE6) hits 68.9% accuracy (AUC 0.714), and 14 benchmarks leak similarly.
• Designed Audit-Prune, a classifier-cleaner algorithm, and open-sourced TruthfulQA-476, cutting audit AUC to near-chance (0.528) while preserving model rankings (Spearman ρ = 0.915).
• Paper: Judging by the Cover: Cleaning Truth Benchmarks to Avoid Surface-Level Feature Leakage — under review.

LLM Activation Steering and Robustness under Dataset Corruption

• Steered verbosity and tone in Llama 3.1 and Qwen-2.5 via residual-stream activation steering (layer/α sweeps).
• Built a guardrailed LLM-judge + JSON-schema evaluation pipeline that reduced hallucinated outputs.
• Made steering resilient to noisy and adversarial dataset corruption via robust high-dimensional mean estimation, in collaboration with Martian AI.
• Released three public contrast datasets on Hugging Face (conciseness–verbosity, positivity–negativity, formal–informal) for steering and evaluation.
• Paper: Understanding and Mitigating Dataset Corruption in LLM Steering — 2026 (arXiv:2603.03206).

Efficient Multi-Dimensional Two-Sample Testing (dKS)

• Extended the Kolmogorov–Smirnov distance to multiple dimensions: a unit-invariant metric (IPM) with a near-linear O(n log n) ε-approximate algorithm.
• Shipped as an open-source C++/pybind11 library — ~76,000× faster than baseline (code, project page).
• Paper: Efficient and Stable Multi-Dimensional Kolmogorov–Smirnov Distance — 2025, under review (arXiv:2504.11299).

MotionPI — Privacy-First Wearable Health Sensing Platform

• Sole developer of an end-to-end mobile/wearable health-sensing system (HRV/PPG wristbands → smartphone app → API → database) for longitudinal, in-the-wild participant data.
• Streamed high-frequency PPG, ENMO/accelerometry, and survey signals with offline-first, schema-validated sync; sustained ~7.7M records/day with zero malformed writes.
• Built an ENMO threshold-calibration visualizer for rapid activity-trigger tuning and data-quality review (code); diagnosed an upstream mobile SDK reliability issue accepted by Flutter maintainers (issue).
• Paper: Designing a Secure and Resilient Distributed Smartphone Participant Data Collection System — EAI SmartSP 2025 (arXiv:2510.19938).

Anomaly Detection — Region-Aggregated Spatial Scan Statistics

• Replaced SaTScan-style centroids with 20–50 points sampled per region: higher detection power, ~3,000× faster than FlexScan on US counties (code).
• Paper: Sampling for Region-Aggregated Spatial Scan Statistics — 2026 (arXiv:2607.01451).

Computer Vision — High-Dimensional Spectral–Spatial Change Detection

• Created a dual-stream 3D/2D CNN with SE attention; accuracy >96%, κ > 0.9, and lower false positives vs. baselines.
• Paper: A Hyperspectral Change Detection Framework Based on Double-Stream CNNs with Attention Module.

Ph.D. in Computer Science

• Advisor: Prof. Jeff Phillips.
• Expected graduation: May 2028.
• GPA: 3.9/4.0
• Focus: LLM interpretability; high-dimensional geometric data analysis.
• Relevant coursework
  • Machine Learning
  • Probabilistic Machine Learning
  • Deep Learning
  • Data Mining
  • Visualization

M.Sc. in Computer Science (Algorithms & Computation)

• GPA: 17.74/20
• Thesis: Graph-theoretic modeling of bushfire propagation.
• Relevant coursework
  • Advanced Algorithms
  • Approximation Algorithms
  • Randomized Algorithms
  • Quantum Algorithms & Computation
  • Graph Algorithms
  • Network Science
  • Internet Algorithms
  • Distributed Systems

Graduate Fellowship

Top-Talent Scholarship (merit)

M.Sc. University Entrance Exam — Rank 29/20,000 (top 0.15%)

Talk — Designing a Secure and Resilient Distributed Smartphone Participant Data Collection System

Poster — Efficient and Stable Multi-Dimensional Kolmogorov–Smirnov Distance (dKS)