I'm an undergraduate student at Smith College, Class of 2027, majoring in Computer Science. I’m passionate about machine learning, data-driven storytelling, and the ethical use of technology to address global challenges.
My work spans healthcare AI, genomic research, and large language models—often with a focus on fairness, accessibility, and real-world impact.
Developed a dermatology classification model for the Break Through Tech x Algorithmic Justice League Kaggle competition.
The model classified 21 skin conditions while addressing bias against darker skin tones.
Key Highlights:
- Used a Vision Transformer (ViT Base Patch 16) with transfer learning
- Integrated Focal Loss and Mixup Augmentation for fairness and generalization
- Applied a variety of data augmentation techniques including CLAHE and noise
- Emphasized explainability using confusion matrix-based visualizations
- Achieved strong results on the Kaggle leaderboard using weighted average F1-score
🧾 Notebook: ViT - skin condition classification.ipynb
📄 Submission: vit_predictions.csv
Researcher at the Veterinary and Biological Informatics Lab at Smith College.
Working on reducing the size and improving the usability of large-scale genomic data.
Goal: Support cancer research by evaluating mice as model organisms for humans.
Investigated prompt tuning techniques for optimizing performance in Large Language Models (LLMs).
Focuses on structure-aware prompts applicable across different tasks and model types.
- Languages: Python, R, SQL
- Libraries: scikit-learn, PyTorch, Pandas, NumPy
- Tools: Jupyter, Git/GitHub, Colab, Kaggle
- Concepts: Transfer Learning, Data Augmentation, Fairness in ML, Prompt Engineering
- 📧 Email: copara@smith.edu
- 🔗 LinkedIn: linkedin.com/in/chiomaopara
I’m currently learning how to:
- Structure data science projects for real-world use
- Contribute to open-source communities
- Apply AI tools ethically across sectors like healthcare, economics, and education
Always open to collaboration and mentorship opportunities!