About Me
Jackson Small
ML Engineer • Healthcare AI Researcher • Data Science @ UCF Honors
Featured Work
Enterprise ML Engineering
Deployed RAG/LLM prototypes and Python automation achieving $700,000 in cost avoidance with 17x faster migration velocity at Insurity. Built HIPAA-compliant audit infrastructure for 23 PHI tables.
View Details →Published Thesis Research
CNN-based EEG seizure detection with Lipschitz stability analysis (L = 24.72). 97% accuracy, 0.99 AUC. Published in UCF STARS Repository with mathematical rigor bridging theory and application.
Read Paper →AI Risk Scoring Platform
End-to-end telematics platform with real-time Streamlit dashboard. 0.98 ROC-AUC, 47 engineered features, gamification system. Complete ML pipeline from simulation to deployment.
View Demo →Impact Metrics
Technical Skills
Core Technologies
Specialized Skills
Sample: Lipschitz Bound Estimation (Thesis Research)
def estimate_lipschitz(model):
"""
Estimates Lipschitz constant of CNN by computing spectral norms
Best model achieved L = 24.72 (B vs. E classification)
"""
norms = []
for layer in model.modules():
if isinstance(layer, (nn.Conv1d, nn.Linear)):
w = layer.weight.detach().cpu().numpy().reshape(layer.weight.shape[0], -1)
s = np.linalg.svd(w, compute_uv=False)
norms.append(np.max(s))
L_hat = np.prod(norms)
return L_hat, norms
View Full Tech Stack →
Languages: Python • SQL • R • JavaScript • Julia • C • Bash • PowerShell
ML/DL Frameworks: PyTorch • TensorFlow • Scikit-learn • SMOTE • Statsmodels
Data Engineering: SPARK • Azure DevOps • AWS • Git • ETL Pipelines
Analytics & BI: Power BI • Streamlit • Plotly • Tableau • Excel
Core Competencies: LLMs/RAG • A/B Testing • Statistical Modeling • Predictive Analytics
Development Tools: Jupyter • VS Code • Docker • Hyprland (Arch Linux)
Skills Proficiency
GitHub Projects
EEG-Seizure-CNN-Thesis
CNN-based EEG seizure detection with Lipschitz stability analysis (L = 24.72). PyTorch implementation with 97% accuracy.
Small_Jackson_TelematicsInsurance
AI-powered driver risk scoring platform with Streamlit dashboard. 0.98 ROC-AUC, 47 engineered features.
shopper-behavior-analysis
E-commerce purchase prediction using K-Means clustering, SMOTE, and Random Forest. 0.93 AUC on 12,330 sessions.
GitHub Activity
Contribution Activity
Top Languages
Certifications
Let’s Connect
Seeking ML Engineering & Healthcare AI Roles
I'm seeking roles where statistical rigor meets production impact. As an Epilepsy Foundation Ambassador, I'm especially passionate about medical AI applications.
Fun Facts
- I run Hyprland on Arch Linux (btw)
- I play drums, cajon, and electric guitar