Hello! My name is Sophia Fente-Damers. I just graduated from The University of Texas at Austin with a Bachelor of Science in Mathematics and a minor in French Studies.
My favorite thing is trying and rating new restaurants on Beli (@sophiafd, give me a follow!). I enjoy laughing out loud and going to comedy shows. I love learning new things and pushing myself out of my comfort zone, like building this website.
I am not a coder by trade, but who is to say you're not something. In this current digital world we have so much information at our fingertips, anything you want to learn you can, anyone you want to be you can become.
I've spent my life in motion, living across three countries and seven cities: from Anchorage to Madrid. This global perspective has taught me to embrace change, a theme that has defined my career path.
I initially entered university with a focus on nursing and medicine, but a single Calculus I course changed everything. I took a leap of faith, switched my major to Mathematics, and never looked back. That decision led me to actuarial internships in Paris and New York City, where I sharpened my analytical skills in the heart of the financial world.
However, my path came full circle during my senior year in Professor Charlee Garden's Health Outcomes & AI Ethics course. Today, I am combining my love for data with my passion for healthcare by developing a machine learning model designed to categorize skin color via the Fitzpatrick Scale. By generating synthetic data and refining classification, my goal is to provide dermatologists with an objective, bedside tool that removes the subjectivity of lighting.
I am currently embarking on my career as a Healthcare Actuary at PwC, staying at the intersection of innovation, risk, and wellness. I am deeply inspired by the startup space and the brilliant ways people solve complex problems. Whether you have a new idea, an opportunity, or just a thought you're working through, let's connect!
I rate everything I eat, follow me on Beli to see my current favorites, always up to date.
A 6-class skin tone classification model trained on ~19,000 dermatological images across three datasets (Fitzpatrick17k, SCIN, PAD-UFES) using Google Vertex AI AutoML Vision, achieving a best PR AUC of 0.774 and 75% precision. Directly addresses the industry-wide dataset skew where over 70% of standard training images represent lighter skin tones. Engineered an augmentation pipeline using flips, rotations, random cropping, and Gaussian noise to address severe class imbalance in Fitzpatrick Types 4–6, targeting a published 27–36% ROC-AUC performance gap that represents real patient risk when AI diagnostic models are deployed on darker skin tones.
2025–2026 · University of Texas at Austin. Developed a comprehensive lithium-ion battery fire risk management report for Lakeshore Recycling Systems, a waste management company processing over 3.5 million tons annually. Designed a four-strategy, defense-in-depth framework addressing LIB fire risk across the full waste lifecycle using a Total Cost of Risk (TCOR) approach, including AI-based visual recognition for battery detection, a YOLOv11 computer vision sorting model, a three-zone suppression framework using Siemens FDA241 detectors and CellBlockEX dry mineral quarantine, and a Source Control framework combining behavioral incentives with Extended Producer Responsibility advocacy. Presented at RISKWORLD in Philadelphia, placing 1st out of 53 global teams and earning a $10,000 prize.
Nov 2025 · University of Texas at Austin. Built a full reinsurance pricing analysis for the CAS Specialty Reinsurance Case Competition, applying catastrophe loss trending, CAT model adjustments, credibility-weighted peril selection, and treaty-level deal economics. Developed trended CAT loss ratios across 2014–2025, calculated experience weights by peril using a structured blending formula, and modeled multiple quota share structures to estimate reinsurer underwriting profit and combined ratio. Demonstrated CloudCover Re could achieve a 25.3% expected margin at a 50% quota share. Delivered a full presentation at the 2025 CAS Annual Meeting after one week of development.
Led the actuarial and strategy workstream for a national health insurance risk case, synthesizing regulatory shifts, demographic trends, and solvency risk into a framework presented to 500+ industry leaders. Won 1st place by audience vote.
Developed a first-of-its-kind market entry strategy for the pension risk transfer space using ALDAs, an approach built independently rather than adapted from existing frameworks. Rebuilt the entire analytical framework mid-competition following a live case revision, under direct scrutiny from Munich Re's senior leadership.
Built from scratch while learning to code. Proof that in this digital world, you can learn anything, and anyone you want to be, you can become.
More coming soon.
I am always open to a conversation, ideas, opportunities, or whatever thought you are working through. Share anything.