CV

Curriculum Vitae

📄 Download 1-Page PDF CV

Professional Profile

AI Researcher specializing in machine reasoning and alignment. Blends formal logic with rigorous low-level systems engineering to architect custom agentic frameworks and highly creative adversarial evaluations. Former Military Intelligence Operator with a Top Secret clearance, bringing a rigorous, threat-modeling mindset to AI security and model capability evaluations.

Technical Skills

• Languages: Python, C++, Java, SQL, LaTeX
• ML & AI Frameworks: PyTorch, vLLM, HuggingFace, Transformers
• Systems & Infrastructure: Linux/Ubuntu Server, AWS, Apache Flink, Distributed GPU Clusters
• Core Competencies: Large Language Models (LLMs), Machine Reasoning, AI Alignment, Context Management, Mechanistic Interpretability, Adversarial Evaluations, Threat Modeling, Distributed Systems

First-Author AI Research

• Robust Reasoning Benchmark (RRB) (2026) Under Review at a top-tier conference | University of Toronto & Vector Institute
  • Designed a highly creative adversarial evaluation framework leveraging 13 deterministic textual perturbations to decouple an LLM’s mechanical deciphering from its underlying mathematical logic.
  • Demonstrated that the well-known phenomenon of attention drift occurs even within a single query’s Chain-of-Thought, empirically showing that intermediate reasoning steps pollute the dense attention mechanism.
  • Engineered custom mechanistic interpretability pipelines in PyTorch to extract and analyze causal attention probability matrices.
  • Identified critical safety-filter vulnerabilities in Claude 4.6 Opus, discovering that current alignment strategies penalize abstract character-level reasoning by misclassifying inputs as prompt injections.
• Context Management Benchmark (2026 – Present) Active Research
  • Designed a custom, tightly coupled LLM agent and evaluation pipline (without MCP solutions) to compile Python/Math tasks into programmable computational graphs, utilizing AST parsing and dynamic rejection sampling to rigorously benchmark multi-step state tracking and working memory. Engineered a targeted repair layer for calculator-augmented agent that enforces tool-use and answer recovery via agent-loop interventions. This improved math benchmark accuracy by +23% to +45% across 3 Qwen models.
  • Testing foundation models’ capacity to manage relevant variables while flushing irrelevant context across long logical trajectories.
• Fusing Adds and Shifts for Efficient Dot Products (2025) IEEE Computer Architecture Letters
  • Proposed and validated a novel hardware algorithmic optimization for dot-product computations.

Engineering & Professional Experience

• Systems & Infrastructure Engineering (MSc Thesis) (2020 – 2022) University of Toronto
  • Engineered a flexible IoT distributed data-streaming framework from scratch, designed to automatically partition computational streaming queries between edge devices and cloud instances.
  • Built the full software stack: programmed Arduino/C++ sensors for real-time biological data collection (EMG/ECG), developed custom socket networking protocols, and deployed cloud infrastructure using AWS and Apache Flink.
• Intelligence Operator (2013 – 2018) Canadian Armed Forces
  • Held a Top Secret security clearance, conducting rigorous analysis of classified information streams to produce actionable intelligence reports for command elements.
  • Developed a strong adversarial threat-modeling mindset, emphasizing operational security, rigorous data validation, and the identification of logical vulnerabilities.
• Mathematics Teacher (2012 – 2013) Blyth Academy
  • Taught foundational mathematics, developing the ability to distill and communicate complex quantitative concepts.

Co-Authored Systems Research

• GPUPool: A Holistic Approach to Fine-Grained GPU Sharing in the Cloud PACT 2022 | Co-authored with Xiaodan Serina Tan, Nandita Vijaykumar, Gennady Pekhimenko.
• Habitat: A Runtime-Based Computational Performance Predictor for Deep Neural Network Training USENIX ATC 2021 | Co-authored with Geoffrey X. Yu, Yubo Gao, Gennady Pekhimenko.

Education

• PhD in Computer Science (Paused to transition to industry), University of Toronto, 2022 – Present
• Master of Science (MSc) in Computer Science, University of Toronto, 2020 – 2022
• Bachelor of Science (BSc) in Computer Science, University of Toronto, 2018 – 2020
• Bachelor of Science (BSc) in Mathematics and Philosophy (Formal Logic), University of Toronto, Graduated 2011