Skip to content
View johnmelel's full-sized avatar
  • Chicago

Highlights

  • Pro

Block or report johnmelel

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don't include any personal information such as legal names or email addresses. Markdown supported. This note will be visible to only you.
Report abuse

Contact GitHub support about this userโ€™s behavior. Learn more about reporting abuse.

Report abuse
johnmelel/README.md

Hi, I'm John Melel ๐Ÿ‘‹

๐ŸŽ“ MS in Applied Data Science @ University of Chicago
๐Ÿง  Specializing in Generative AI, NLP, Bayesian Modeling, and Machine Learning
๐Ÿ“ Based in Chicago, IL
๐Ÿ”— LinkedIn | ๐ŸŒ GitHub


๐Ÿงญ About Me

I am a data scientist passionate about building intelligent systems at the intersection of healthcare, machine learning, and generative AI. With a solid foundation in statistics, deep learning, and NLP, I thrive in complex problem spaces and enjoy pushing the boundaries of whatโ€™s possible with modern AI.

My experience at ZS Associates, a global healthcare consulting firm, equipped me to excel in high-pressure, fast-paced environments. There, I led machine learning initiatives, streamlined analytics pipelines, and mentored new team membersโ€”all while delivering impactful solutions for top pharmaceutical clients.

Currently pursuing my Masterโ€™s at the University of Chicago, Iโ€™m actively working on a capstone project that applies LLMs and agentic frameworks to real-world clinical settings, combining multimodal retrieval-augmented generation (RAG) and reinforcement learning.


๐Ÿ”ฌ Key Projects

  • Designing a centralized vector store with multimodal RAG and agent-based reasoning to enhance clinical decision support at UChicago Medicine
  • Leveraging frameworks like AutoGen, CrewAI, and LangGraph to implement multi-agent LLM pipelines
  • Extracted ideal patient cohorts for clinical trials from medical data and doctorsโ€™ notes using NLP pipelines and PU classification
  • Helped simulate patient inclusion strategies to enhance clinical trial success with faster patient identification
  • Trained a C-VAE model to simulate age progression from facial portraits
  • Integrated custom encoder-decoder architecture with PyTorch
  • Developed a deep learning-based LSTM model to forecast London's Air Quality Index (AQI) using over 10 years of hourly data
  • Benchmarked performance across multiple models: ARIMA, SARIMA, ARFIMA, ETS, BSTS, and LSTM
  • Achieved best performance with LSTM (RMSE = 10.2), significantly outperforming traditional statistical approaches

๐Ÿ›  Tech Stack

Languages & Tools

Libraries & Frameworks

Cloud & Platforms


John's GitHub stats


๐Ÿ“ˆ Let's Connect

I'm always open to collaboration, mentorship, and exploring new frontiers in data science. If you're working on something exciting in LLMs, agent systems, or healthcare AI, feel free to reach out!

๐Ÿ“ซ johnjojimelel@gmail.com
๐Ÿ”— LinkedIn
๐ŸŒ GitHub

Pinned Loading

  1. TimeSeries_AQI_Forecasting TimeSeries_AQI_Forecasting Public

    Forecasting London's Air Quality Index using advanced time series methods for our final group project

    HTML 1

  2. bhstoller/AgeTransform-VAE bhstoller/AgeTransform-VAE Public

    Bayesian Machine Learning & Generative AI Final Project

    Jupyter Notebook 2 2

  3. MachineLearning1_CT MachineLearning1_CT Public

    Jupyter Notebook 2 1

  4. coursera coursera Public

    Jupyter Notebook

  5. simulation-codes simulation-codes Public

    Python

  6. copilot-codespaces-vscode copilot-codespaces-vscode Public template

    Forked from github-education-experiences/copilot-codespaces-vscode

    Develop with AI-powered code suggestions using GitHub Copilot and VS Code