• Engineered an adaptive semantic LLM cascade router directing user queries to cost-optimal models based on complexity and intent, slashing inference costs by 35%.
• Pre-trained an edge-capable small language model (SLM) leveraging a Tiny Recursive Model (TRM) architecture with recursive weight-tying, cutting memory footprint by 50%.
• Investigated out-of-distribution generalization in Supervised Fine-Tuning (SFT) to mitigate the knowing-to-doing gap and enhance instruction-following fidelity.
• Delivered technical seminar and fine-tuning workshops on LLM adaptation and LoRA workflows for 250+ enterprise engineers.

• Post-trained Transformers (LLaMA, BERT) on long-context clinical narratives using PyTorch, Hugging Face, and DDP across an 8-GPU Slurm cluster, achieving record 0.70 R² score.
• Architected agentic inference engines leveraging vLLM for continuous batching and PagedAttention, orchestrating hierarchical planning and tool calling.
• Deployed containerized local LLM services (Docker, Ollama/vLLM) supporting 10,000+ university users with sub-second TTFT and optimized KV caching.
• Trained conditional diffusion models on Slurm HPC clusters to sample continuous molecular dynamics and protein trajectories (published in ACS JCTC).
• Co-authored 6 peer-reviewed papers spanning AAAI, NAACL NLP4DH, APA JOPACS, Frontiers in AI, and ACS JCTC.

• Analyzed large-scale scientific peer-review corpora across premier AI/ML venues.
• Built NLP processing pipelines with DistilBERT and fine-tuned classifiers to evaluate review depth and detect automated/synthetic text artifacts.
• Investigated longitudinal review quality trends to assess empirical impacts of LLMs on scientific peer evaluation.

• Instructed 100+ undergraduate and graduate students in ML theory, gradient descent, and transformer attention mechanisms; led weekly labs and office hours.
• Mentored students in implementing ML algorithms, backpropagation, and deep learning architectures from scratch in PyTorch and scikit-learn.
• Designed laboratory curriculum in digital systems and VLSI testing, guiding students on JTAG boundary-scan controllers and Verilog HDL on Xilinx FPGAs.

• Designed and delivered an intensive 4-week Machine Learning course for high-achieving high school students, covering neural networks, intelligent agents, search algorithms, and data preprocessing.

• Awarded the Robert Mayer Undergraduate Research Fellowship to architect an agentic recommender system for degree planning using GraphRAG (Neo4j) and vector search (Qdrant).
• Published and presented at AAAI 2025 Spring Symposium (Human-Compatible AI) and NCUR 2025.

GPA: 3.96 / 4.0

Master’s Thesis: Predictive Natural Language Metrics of Alzheimer’s Disease and Cognitive Decline Trend Analysis (supervised by Dr. Jennifer Dworak, Dr. Eric Larson, and Dr. Jia Zhang).

Key Coursework: Machine Learning in Python, Algorithm Engineering, Computer Architecture, Quantum Computing.

GPA: 3.73 / 4.0 (Triple Major)

Honors: Robert Mayer Undergraduate Research Fellow, Carrie and Edwin Mouzon Mathematics Scholarship, SMU Discovery & Distinguished Scholar.