cv

LLM Activation Steering with Guardrailed Evaluation

• Controlled verbosity and tone in Llama 3.1 and Qwen-2.5 via residual-stream injections (layer/α sweeps).
• Built judge-driven evaluation with JSON-schema checks; reduced hallucinations and stabilized behavior.

dKS — Efficient Multidimensional Two-Sample Testing

• Near-linear 2D algorithm (O(n log n)); ε-accurate, unit-invariant IPM; stability gains vs. mdKS.
• Paper: Efficient and Stable Multidimensional Kolmogorov–Smirnov Distance (SIMODS submission; arXiv: 2504.11299).

MotionPI — Privacy-First Wearable Sensing (Full-Stack)

• Flutter + BLE wristbands → Node.js/Express + MongoDB; background pipelines and secure ingestion.
• Deploy in Huntsman Cancer Institude.
• Interactive visualization for activity-trigger thresholds.
• Paper: Designing a Secure Distributed Participant Data Collection System — EAI SmartSP 2025 (accepted).

Region-Aggregated Spatial Scan Statistics

• Replaced centroid scans with multi-point region sampling; increased detection power; C++ experiments.
• Under review - double blind.

Ph.D. in Computer Science

• Advisor: Prof. Jeff Phillips.
• Focus: LLM interpretability; geometric data analysis; spatial statistics.
• Relevant coursework
  • Machine Learning
  • Probabilistic Machine Learning
  • Deep Learning
  • Data Mining

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

• GPA: 17.8/20
• Thesis: Graph-theoretic modeling of 2020 Australian bushfire propagation.
• Relevant coursework
  • Advanced Algorithms
  • Approximation Algorithms
  • Randomized Algorithms
  • Quantum Algorithms & Computation
  • Graph & Network Algorithms
  • Internet Algorithms
  • Distributed Systems

Graduate Fellowship

Top-Talent Scholarship (merit)