Ignacio Villasmil
Development
Florida, United States
Skills
Machine Learning (ML)
About
Ignacio Villasmil's skills align with IT R&D Professionals (Information and Communication Technology). Ignacio also has skills associated with Database Specialists (Information and Communication Technology). Ignacio Villasmil appears to be an entry-level candidate, with 23 months of experience.
View more
Work Experience
Machine Learning (NLP) Intern
Boston Scientific
May 2023 - August 2023
- Researched and designed proof of concept NLP algorithm for free-text responses from chronic-pain patients under spinal cord stimulation (SCS) for assessment of patient states (Python). Designed preprocessing & feature engineering pipeline of patient free-text responses (via Transformer models), involving filtering of relevant features found in text (Semantic Search Sentence Transformer) and text classification (BART MNLI Transformer). Developed a KMeans clustering model (with optimal number of clusters via silhouette scores & the elbow method) for assessment of chronic-pain SCS patient states using the engineered feature set, with accuracies of 81% and 76% for the two resulting cluster states. Designed interactive and user-friendly LLM chat bot (via prompt engineering of system prompt and feature-specific prompts using Meta's Llama 2) for enhanced feature acquisition and improved model performance from patient text.
Machine Learning Engineer
Center of Neuroengineering & Therapeutics (CNT)
May 2022 - Present
- Worked as part of the Pioneer Project, focusing on analyzing signals from wearable devices (i.e. Apple Watch heart rate & accelerometer) to preemptively detect oncoming seizures in EMU epilepsy patients (Python). Designed preprocessing pipeline for wearable device data cleaning (including moving averages for signal smoothing & synchronization of sampling rates of different signals) & feature engineering (feature selection & scaling) Created a variety of unsupervised learning clustering models, such as KMeans, Agglomerative Hierarchical Clustering, and Gaussian Mixture Model (with optimal number of clusters via silhouette scores & Bayesian information criterion), for the detection of a patient's medical states over several days.
Education
University of Pennsylvania
MSE in Data Science
