Founder, LOGARITHM LLC · AI/ML Engineer · Applied Research & Software
I build software that turns messy data and manual workflows into systems people can use, test, and improve.
I’m the founder of LOGARITHM LLC, where I work across custom software, data engineering, machine learning, and workflow automation. I graduated from Salisbury University with a double major in Computer Science and Data Science.
My interests connect practical engineering with deeper questions: How do we search for useful model structures? How do we distinguish a promising result from overfitting? How can an intelligent system make its decisions easier to inspect?
- LOGARITHM: Growing a technical consultancy around problems in healthcare analytics, operational planning, and applied AI.
- Clinical simulation: Developing Rounds with my team, with a focus on the backend, controlled patient responses, student evaluation, and adaptive practice.
- Healthcare data workflows: Scoping automation for physiological signal processing, alignment, visualization, and analysis.
- Search and learning systems: Exploring symbolic regression, program synthesis, equation learning, and search-guided model discovery.
Undergraduate research · Symbolic regression & financial time series
Developed a Python research pipeline that generates, optimizes, and evaluates candidate model structures using genetic programming and grammar-guided search. The work combines NumPy/Numba computation, time-series validation, Monte Carlo permutation testing, and ensemble analysis.
This research shaped how I approach ML: a strong fit is a starting point; robustness, search bias, and the structure of the solution space deserve equal attention.
Public structural-regression experiments
2nd Place — TwinVerse Track, Rutgers Health Hack 2026
An early-stage virtual-patient platform for practicing clinical interviews. I built the backend and AI evaluation pipeline, combining faculty-defined patient information, dialogue-action classification, LLM-based encounter evaluation, and evidence-linked debriefing.
The prototype incorporated 8 patient profiles, 534 source-linked clinical facts, and all 7 encounter-rubric domains. Ongoing work explores how student-specific feedback can guide the next practice scenario.
Operational software · MVP in development
Building a scheduling tool around employee availability, time off, position hierarchies, historical schedules, and staffing requirements. The system explores both assigning strong fits and developing staff capability over time, with import workflows for existing scheduling data.
Computer vision · Software engineering capstone
Worked with a team on automated chick counting using Ultralytics YOLO, line-crossing logic, and downstream models built from detection features. My work focused on turning detector outputs into more reliable counts and evaluating error across counting intervals.
1st Place — Women’s Health Track, Rutgers Health Hack 2025
Developed the core AI/ML logic for a team-built OB/GYN triage prototype, using text classification and guided follow-up questions to refine patient intent and support subspecialty routing.
| Area | Tools & methods |
|---|---|
| ML & numerical computing | Python, NumPy, pandas, SciPy, scikit-learn, PyTorch, TensorFlow/Keras, Numba |
| Search & evaluation | Genetic programming, symbolic regression, Monte Carlo methods, time-series validation, ensemble modeling |
| Vision & language | Ultralytics YOLO, OpenCV, TF-IDF, logistic regression, LLM integration and evaluation |
| Software & data | SQL/SQLite, JavaScript, React, Next.js, Java, C/C++, Git, Linux/WSL, CUDA |
I also share experiments in ARC-AGI program search and my ML sandbox.
- Former AI Club President and Computer Science tutor/lab assistant at Salisbury University.
- NCAA All-American in discus, three-time NCAA championship qualifier, school-record holder, and multi-time conference champion.
- Experience in restaurant operations that informs how I think about staffing, workflows, and useful software.
- I also love powerlifting, my bests are: 445 Squat, 370 Bench, 535 Deadlift @ 215 BW
Much of my ongoing research and business work lives in private repositories. This profile contains a mix of public experiments, team projects, and earlier coursework.
For project inquiries or technical collaboration, reach out through LOGARITHM, email, LinkedIn.