MLE / Data Science / AI for Science
M.S. Data Science student at UC San Diego bridging domain expertise with modern AI. Focused on machine learning engineering, data pipelines, and robust evaluation systems for complex real-world and scientific datasets.
I am positioning my work around a practical question: how can modern machine learning help traditional engineering and scientific fields move faster, measure better, and make better decisions from complex data?
M.S. Data Science student with a physics and materials-science research background, now focused on machine learning engineering, scientific data systems, and data science workflows that connect experiments, models, and decisions.
I like systems that can be defended: reproducible pipelines, honest validation, clear assumptions, and interfaces that let scientists, engineers, and product teams understand model behavior before relying on it.
I work across the modeling stack: data pipelines, statistical reasoning, ML evaluation, and domain-aware analysis for product and scientific datasets.
> define_problem() metric = "scientific decision quality" constraints = ["noise", "bias", "domain shift"] > build_pipeline() steps = [ "clean experimental data", "engineer physics-aware features", "train and validate models", "ship reproducible inference" ] > ship_insight() output = "model + uncertainty + domain context"
Modeling pipelines, evaluation loops, feature stores, batch inference, and deployment-minded engineering for real-world data products.
Feature engineering and model validation for noisy measurements, spectra, microscopy, diffraction, energy systems, and materials workflows.
Power analysis, KPI design, uncertainty-aware comparison, and careful claims when moving from measured effects to decisions.
Interfaces for communicating model diagnostics, metric narratives, scientific context, and decision-ready summaries to mixed audiences.
Selected work across experimentation, signal modeling, and NLP highlights practical modeling, evaluation, and communication skills.
Designed A/B test workflows, power analysis, and statistical evaluation patterns for deciding whether a product or modeling change is truly moving the target metric.
Built a reusable mental model for turning noisy measurements into features, validating signal quality, and communicating uncertainty before a model is used downstream.
Built an end-to-end NLP pipeline fine-tuning DistilBERT for extreme class imbalance, improving rare-event detection F1 to 0.84 and reducing false positives in simulation.
Selected publications connect my earlier work in materials, energy systems, semiconductors, spectroscopy, and computational imaging with my current focus on machine learning and scientific data systems.
A foundation in physical-science research informs a practical approach to AI: respect the measurement process, model uncertainty, and build tools that scale scientific decision-making.
Advanced Energy Materials, 2022. Work on durable energy devices, catalyst design, and performance measurement under real operating constraints.
NIST-linked materials measurement work demonstrating data fluency: analyzed complex X-ray diffraction patterns, bandgap measurements, and structure-property relationships to compile reproducible reference data.
Co-authored review on semiconductor and spintronic devices, summarizing radiation environments, MgO tunnel barriers, and microstructural evidence for robust electronics.
Investigated complex phase relationships and structured scientific datasets, laying the groundwork for incorporating thermodynamic constraints and domain knowledge into machine learning models.
Developed computational imaging algorithms for signal recovery and image reconstruction, bridging classical optics with modern signal processing and machine learning workflows.
The timeline connects prior research training with current goals in machine learning engineering and data science.
University of California San Diego
Georgia Institute of Technology. Conducted research on experimental data analysis.
National Institute of Standards and Technology. Assisted with physical sciences research and data collection.
Delaware State University