CF Crispin Foli Biomedical Engineering | Data Science | ML
“

About

I am a 5th-year biomedical engineering PhD candidate at Case Western Reserve University specializing in Brain-Computer Interface (BCI) systems. As part of the Laboratory for Intelligent Machine-Brain Systems (LIMBS), I develop end-to-end machine learning and signal processing pipelines to decode speech and movement intentions from high-density neural recordings across multiple brain regions and human participants.

Beyond my doctoral research, I enjoy building deep learning systems, particularly involving multi-modal learning, computer vision, and real-time model deployment across a wide variety of applications (check out my portfolio).

In my spare time, I enjoy watching movies and sports, and playing FIFA (some say I'm one of the best!).

Open to Work Looking for roles as a Data Scientist, ML Engineer, Applied/Research Scientist.
View Projects Get in Touch
Portrait of Crispin Foli Black and white portrait of Crispin Foli
LinkedIn GitHub Google Scholar CV

Education

  • PhD in Biomedical Engineering

    Case Western Reserve University · Cleveland, OH · 2021 – 2026

  • Master of Science in Biomedical Engineering

    Arizona State University · Tempe, AZ · 2019 – 2021

  • Bachelor of Science in Biomedical Engineering

    Kwame Nkrumah University of Science & Technology · Kumasi, Ghana · 2016 – 2020

Expertise

A blend of neural engineering, machine learning and data science.

Programming

Building research-grade and production-ready systems.

  • Python
  • SQL
  • MATLAB

Machine Learning

Classical ML for high-dimensional, noisy scientific data.

  • Clustering (KMeans, hierarchical, etc.)
  • Dimensionality reduction (PCA, t-SNE)
  • Ensemble modeling (Random Forests, stacking)
  • Hyper-parameter optimization (Grid/Random Search)

Deep Learning

Designing and fine-tuning modern neural architectures.

  • Multi-modal learning & computer vision
  • Transfer learning & regularization
  • ViTs, VLMs, CNNs, RNNs (LSTMs, GRUs)
  • PyTorch
  • TensorFlow
  • Keras

Model Deployment & MLOps

Taking models from prototype to reliable services.

  • GCP
  • AWS
  • Hugging Face
  • Docker
  • FastAPI
  • Streamlit
  • Gradio

Research

Data-driven experimentation at the intersection of neuroscience, engineering, and machine learning.

  • Experimental design
  • Scientific writing
  • Brain-Computer Interfaces (BCI)

Data Analysis & Signal Processing

Extracting robust structure from complex time series data.

  • Time & frequency-domain analysis
  • Digital filtering
  • Statistical inference & hypothesis testing
  • Predictive modeling & visualization

Projects

Selected work at the intersection of healthcare, AI, and neuroscience.

ChestVision-PRO interface
Medical AI · Multimodal

ChestVision-PRO

A multimodal deep learning system that combines Vision Transformers with vision-language models to detect diseases and generate clinical-style explanations from chest X-rays.

  • ViT-based multi-label classifiers for co-occurring pathologies
  • Integrated VLMs for post-hoc, human-readable explanations
  • Designed for scalable deployment via Hugging Face Spaces
  • API-ready for integration into assistive diagnostic workflows
Multi-modal Transformer ViT VLM PyTorch Medical AI
Visit ChestVisionPRO →
EMG-based Hand Gesture Recognition spectrogram
EMG · ML

EMG-based Hand Gesture Recognition

Highly optimized lightweight models for decoding hand gestures from multi-channel electromyography (EMG) signals

  • SVM and ensemble models (Random Forests, XGBoost, AdaBoost) trained on multi-channel EMG signals
  • >0.99 test accuracy and F1 scores on held-out gesture classification tasks
  • Includes a simulated real-time inference pipeline for fast and accurate gesture detection
Random Forest XGBoost SVM Scikit-learn GridSearch EMG
Visit EMG-based Gesture Recognition →
Machine-learning toolbox for neural decoding project summary
BCI · Neural Decoding

Machine-learning toolbox for neural decoding

A modular toolkit for decoding movement kinematics from high-dimensional brain signals.

  • Framework for training and benchmarking models for neural decoding
  • Supported architectures: XGBoost, Feedforward Neural Networks, LSTM decoders, Hybrid CNN-MLP models, Kalman Filters, and Ridge Regression
  • Includes example pipelines for preprocessing neural data, training models, and evaluating decoding performance
  • Adapted from Glaser, J.I., Chowdhury, R.H., Perich, M.G., Miller, L.E., & Kording, K.P. (2017). Machine Learning for Neural Decoding. eNeuro, 7.
LSTM CNN Ridge Regression Scikit-learn GridSearch BCI Neural Decoding
Visit Kinematic-Decoding-4-BCI-Control →
ML for Financial Forecasting and Risk Analysis project summary
Finance · ML

ML for Financial Forecasting and Risk Analysis

Applications of machine learning for financial prediction and risk analysis across multiple real-world datasets

  • Regression and forecasting models for housing price prediction
  • Lending risk modeling with classification and class-imbalance handling (minority oversampling)
  • Social media sentiment analysis for financial signal extraction
  • Models include CNNs, SVMs, Random Forests, and Gradient Boosting with extensive feature engineering and hyperparameter tuning
Natural Language Processing Feature Engineering XGBoost CNN GridSearch
Visit AI-4-Financial-Modeling →
Vision-Language

Vision-Language Models for Zero-Shot Classification of Homophones

A simple tutorial on zero-shot image classification, with a particular focus on classifying visual representations of homophones (e.g., mouse (electronic) vs. mouse (mammal)).

  • CLIP-based vision-language models for accurate zero-shot classification without task-specific training
  • Illustrates how visual context resolves semantic ambiguity in homophones (words with identical pronunciation)
  • Interactive demo deployed on Hugging Face Spaces
Multi-modal Transformer VLM Zero-shot Classification
Visit SemanticVision →
ChestVision interface
Medical AI

ChestVision

A multi-label medical diagnosis platform providing access to fine-tuned CNN-based classifiers for automated chest X-ray analysis, detecting multiple thoracic conditions simultaneously.

  • Transfer learning with modern ConvNet backbones (ConvNeXt, EfficientNet, etc.)
  • Optimized for multi-label classification with BCEWithLogitsLoss
  • Evaluated with AUROC, F1-score, and mAP
  • Built for fast, efficient inference in clinical-style settings
CNN Multi-label PyTorch Medical AI
Visit ChestVision →

Certifications

Advanced Data Analytics Professional (Google) certificate

Advanced Data Analytics Professional (Google)

Google on Coursera · End-to-end data analytics specialization.

Verify at coursera.org →
Deep Learning and Neural Networks with Keras (IBM) certificate

Deep Learning and Neural Networks with Keras (IBM)

IBM on Coursera · Keras-based deep learning fundamentals

Verify at coursera.org →

Publications

Neural Mechanisms of Mixed Speech and Grasp Representation figure

Neural Mechanisms of Mixed Speech and Grasp Representation in Sensorimotor Cortices

Foli, C., Conlan, E.C., Memberg, W.D., Bhat, P., Graczyk, E.L., Johnson, T.R., Taylor, D.M., Herring, E.Z., Sweet, J.A., & Ajiboye, A.B.

bioRxiv · 2026

Targeting Optimal Grasp-Related Cortical Areas figure

Targeting Optimal Grasp-Related Cortical Areas for Intracortical Brain-Machine Interfaces after Spinal Cord Injury

Johnson, T.R., Foli, C., Conlan, E.C., Koenig, K.A., Lowe, M., Memberg, W.D., Kirsch, R.F., Herring, E., Bazarek, S.F., Graczyk, E.L., Taylor, D.M., Ajiboye, A.B., & Sweet, J.

medRxiv · 2025

Cortical representation of sensation figure

Cortical representation of sensation elicited by peripheral nerve stimulation in an individual with incomplete spinal cord injury

Bhat, P.R., Memberg, W.D., Hutchison, B., Spilker, B.J., Bose, R., Foli, C., Ketting-Olivier, A., Taylor, D.M., Kirsch, R.F., Herring, E., Sweet, J., Miller, J.P., Ajiboye, A.B., & Graczyk, E.L.

medRxiv · 2025

Contact

Interested in research collaborations, consulting, or speaking opportunities? Let’s connect.

Email cxf306@case.edu
LinkedIn linkedin.com/in/cfoli77
GitHub github.com/cfoli
Google Scholar View profile
✓ Email copied to clipboard!