I'm a passionate Data Engineer with a focus on Natural Language Processing. I work on building data and machine learning systems. I love tackling real-world problems, from designing Healthcare AI data pipelines to developing ML models that can classify cognitive thought patterns from text. These efforts can lead to better health outcomes. I enjoy working across different data areas and have hands-on experience in healthcare, geospatial, and language based datasets. Outside of work, I’m interested in psychology and fashion.

• 🔭 Currently working as a Data Scientist at John Snow Labs, developing Spark based healthcare AI pipelines that harness NLP and deep learning to improve clinical insights and patient outcomes.
• 🌱 Actively learning about AI Agents & Generative AI
• 💡 Passionate about applying ML to healthcare, natural language, and social good
• 👯 Always open to collaborating on meaningful, impactful projects
• 🎨 My hobbies are painting, reading and creating digital art.
• 📫 Reach me at paulamib123.dev@gmail.com

• 📊 Netflix Analytics Data Pipeline — End-to-End Data Engineering Pipeline

  Built an end-to-end Netflix analytics pipeline to ingest, transform, and analyze streaming data using automated ETL workflows, SQL-based analytics, and data visualization for actionable insights.

• 📚 BookReview App — Full Stack Web App

  Created a full-stack book review platform with authentication, search, and review features using React frontend, Flask REST API backend, PostgreSQL database, and Heroku deployment.

• 🧠 JournalWise — Mental Health based ML Pipeline

  Built a full-stack ML app to detect negative thought patterns from journal entries using Reddit-scraped mental health data, Logistic Regression/SVM models with augmentation, FastAPI + Streamlit deployment, and Gemini API for positive thought reframing.

• 🧬 DNA Language Model — Cross-Species Codon Optimization

  Developed a transformer-based DNA sequence generator using extended CodonBERT with amino acid/species embeddings, trained on 337K+ sequences across 5 organisms achieving a test perplexity of 8.52.

• 🛡️ Toxic Comment Classifier — Content Moderation with NLP

  Built a multi-label toxicity classifier on 172K+ Wikipedia comments, improving ROC-AUC from 0.71 (Logistic Regression baseline) to 0.94 using fine-tuned DistilBERT with detailed bias/error analysis.

"Engineering is the closest thing to magic that exists in the world.✨"