Victor P.

AI Model Evaluation:

May 2026 - Present • 2 mos

Perform red-teaming and comparative performance analysis on advanced coding agents and LLMs to optimize developer workflows and output reliability. Designing and implementing comprehensive rubrics and defining key performance indicators to quantify model accuracy and optimize output reliability.

Lead Software Engineer

Aug 2024 - Present • 1 yr 11 mos

Architecting an AI-powered medical bill auditing platform using Node.js and Google Gemini to automate the detection of billing errors and insurance inconsistencies. Implementing RAG (Retrieval-Augmented Generation) principles to analyze complex medical coding against a database ingested from CMS datasets and public fee schedules as a cost-effective approach. Designed a secure, scalable backend using Supabase for data persistence and Google OAuth 2.0 for HIPAA-compliant authentication frameworks. Integrated Stripe API for automated subscription billing and secure transaction workflows.

Independent Technical Consultant

Sep 2018 - Present • 7 yrs 10 mos

Developed custom Python-based quantitative trading tools utilizing Pandas and NumPy to backtest technical analysis strategies in the equities market. Used Monte Carlo simulation to evaluate the 'Family Glitch' policy fix, utilizing a synthetic population of 50,000 households to model insurance migration trends. By applying nuanced statistical distributions, including a triangular distribution skewed toward the 250% FPL 'subsidy sweet spot' - I identified the precise affordability thresholds where employer-sponsored plans become economically non-viable. This analysis provided actionable intelligence for market prediction, identifying a ~12% eligibility change that traditional 'average-based' modeling would fail to capture.