Projects

Machine learning from first principles, reliable classification pipelines, and structural engineering.

Neural Network from Scratch (MNIST Digit Identifier)

Python • NumPy
Multilayer Perceptron & Hand-Derived Backpropagation
MNIST handwritten digit classification demo showing activation probabilities
Interactive inspector: classifying test digit '2' alongside output layer activation probabilities (99%+ confidence)
  • Engineered and implemented a multilayer perceptron (784-16-16-10 structure, ~13K parameters) from scratch using NumPy with no machine learning libraries.
  • Derived analytical matrix gradients by hand to train with a custom stochastic gradient descent (SGD) loop, achieving ~94.2% test accuracy.
  • Built a binary IDX parser with Python’s struct module to unpack raw MNIST bytes into normalized arrays.
  • Profiled training bottlenecks between full-batch and stochastic gradient descent, discovering parameter update cadence to be the critical factor for rapid convergence.
Python NumPy Matplotlib Neural Networks Backpropagation SGD Optimization
View on GitHub ↗

End-to-End Tabular Classification Pipeline (Titanic)

2026
Custom feature engineering & leakage-safe model evaluation
  • Built a Scikit-Learn classification pipeline with fold-isolated preprocessing to prevent data leakage.
  • Implemented custom transformers for title, family, and cabin features, with conditional age imputation.
  • Compared Logistic Regression, Random Forest, and Gradient Boosting using GridSearchCV, achieving 0.894 cross-validation ROC-AUC.
PythonScikit-LearnPandasNumPyGridSearchCV
View on GitHub ↗

Deck Arch Truss Bridge Design & Optimization

2023 – 2025
Delaware Bridge Design Competition • 1st Place State Champions
Complete AutoCAD technical blueprint: plan, elevation, section A-A, and isometric views with full dimensioning
Complete AutoCAD technical blueprint: plan, elevation, section A-A, and isometric views with full dimensioning (Delaware Bridge Design Competition • Team: The Peculiar Pythons • Rev 8)
  • Designed, simulated, and assembled optimized balsa wood bridges across three years of iterative competition.
  • Constructed 3D beam member assemblies and performed stress and displacement simulations using SOLIDWORKS.
  • Calibrated finite element models using empirical scaling factors (2.52–3.02) to match real-world physical load tests.
  • Iterated through 8 distinct CAD revisions and manufactured a monolithic steam-bent curved arch to eliminate joint weaknesses.
SOLIDWORKS FEA AutoCAD 2D Drafting Empirical Calibration Structural Prototyping
Read Engineering Case Study →

Let's connect

Email Address
bryanmsh@gmail.com
Location
Amherst, MA / Boston, MA