Jan 2026 – May 2026

Machine Learning Engineering Intern

Rolls-Royce — Remote

  • Built a Python framework for generating synthetic turbofan telemetry datasets, enabling PHM researchers to train and benchmark models without accessing ITAR-restricted engine data.
  • Developed a GPU-accelerated conditional diffusion pipeline with transformer-based masked denoisers and inpainting-style sampling, scaling generation to 125k+ telemetry samples per run (~50× throughput increase).
  • Implemented column-agnostic ingestion, dynamic schema resolution, multi-file CSV loading, LoRA fine-tuning, and synthetic-data evaluation with MMD/domain-classifier metrics across NASA and Rolls-Royce datasets.
  • Worked on Rolls-Royce-sponsored engineering initiatives through Purdue's applied industry collaboration program.
Python PyTorch CUDA Diffusion Models LoRA
Sept 2025 – Present

Software Engineering Intern

AbbVie — Remote

  • Built an LLM reliability autograder with multi-epoch execution and full reproducibility, cutting evaluation noise 40%+ on complex statistical tasks.
  • Designed an error taxonomy and agreement scoring for the autograder, boosting grader consensus 35% and surfacing 20% more actionable failure patterns.
  • Optimized cross-architecture autograder evaluation workflows, delivering 50%+ cost savings and 8× faster iteration cycles for production benchmarking.
Python LLMs Ellmer Vitals
Aug 2024 – July 2025

Algorithms Lead

Purdue Robomasters — West Lafayette, IN

  • Led omni-directional motion algorithms in ROS, fusing forward/inverse kinematics and real-time feedback in Gazebo for precise holonomic control in RoboMaster competition.
  • Tuned PID controllers via Gazebo sim-to-real in ROS, cutting latency/overshoot 50%+ to improve chassis stability and battle responsiveness.
  • Built CV-based autonomous aiming in ROS integrating tracking algorithms, boosting targeting precision 35% and robustness under arena conditions.
ROS Gazebo OpenCV C++ Control Systems