About

Learn More About Me

Hi, I'm Etsub Wendwosen Feleke, a recent Data Science graduate who is passionate about NLP, LLMs, and agentic AI systems. I hold a B.A. in Economics (4.0 GPA) and an M.S. in Data Science from the University of North Texas (4.0 GPA). This site is where I share the projects, experiments, and small victories from my ongoing journey in data science and AI.

I have experience applying machine learning and data analysis to real problems in global health and financial consulting, from building forecasting models on large healthcare datasets to designing dashboards and pipelines for a consulting and microfinance firm. I'm currently learning more about agentic AI, multi-agent systems, retrieval augmented generation, and the Model Context Protocol (MCP), and how these pieces fit together to build reliable, production ready AI tools.

I'm currently interested in Data Scientist and ML Engineer roles, especially at mission driven organizations working in health, education, or government, where I can apply AI and NLP to problems that matter.

What excites me most is building agentic AI systems, evaluation frameworks, and production grade ML solutions, along with the Model Context Protocol (MCP) and everything it opens up for connecting AI models to real tools and data. I'm the kind of person who ends up down a rabbit hole reading documentation for fun, then wants to build something with it by the weekend.

A Little About My Experience:

• I worked as a Machine Learning Engineer at JSI (John Snow, Inc.), where I helped build forecasting models on large healthcare datasets and automated pipelines to support global health programs.
• Before that, I worked as a Data Analyst at Glocal Management Partners, a consulting and microfinance firm, where I built ETL pipelines and dashboards, and used time series forecasting to support investment decisions.

Tech Stack I Work With:

• AI & LLMs: LLM Integration (GPT-4, Claude), Agentic AI, Prompt Engineering, RAG, Vector Databases (ChromaDB, FAISS), MCP Server Development (FastMCP), Multi-Agent Systems (LangGraph)
• Machine Learning & Deep Learning: PyTorch, TensorFlow, Scikit-learn, CLIP, ResNet, Computer Vision, Predictive Modeling, Time Series Forecasting (ARIMA), Random Forest, OLS/Ridge/Lasso Regression, SHAP Explainability
• NLP & Text Mining: Text Classification, Embeddings, Semantic Search, Sentiment Analysis
• Data Engineering & MLOps: ETL Pipelines, Airflow, MLflow, Azure, Google Cloud Run, Docker
• Visualization & BI: Tableau, Power BI, Matplotlib, Recharts
• Web & Deployment: React, FastAPI, Flask, Serverless (Cloud Run)
• Languages & Tools: Python, SQL, R, Git, REST APIs, Bash

Currently Focusing On:

• Researching human in the loop design for multi-agent systems, through my project RedInk, a multi-agent AI grading assistant built with LangGraph that keeps a human reviewer in the loop before any grade is finalized.
• Studying how AI models can be given persistent memory and tool access through the Model Context Protocol, through my project Cognitive Canvas, a custom MCP server built for Claude and GitHub Copilot.
• Studying the relationship between AI adoption and firm performance, through Agentic Evolution, an empirical analysis of AI adoption and revenue growth among S&P 500 firms using panel data and explainable ML.
• Researching content moderation methods for near duplicate detection, through Deja Vu, a study using CLIP and ResNet-50 to identify modified re-uploads of banned video content.

A Bit More About Me:

• I'm a curious person at heart, and that curiosity is what pulled me from economics into data science in the first place. I get genuinely excited switching domains. My work has spanned healthcare, microfinance, content moderation, educational technology, and empirical economics, and each new domain has made me a better data scientist, not a more scattered one.
• I love collaborating with others to tackle real world problems. Some of my favorite moments in this field have come from bouncing ideas off teammates until a messy problem finally clicks into place.
• Mentorship matters a lot to me. I remember what it felt like to be new to this field and unsure where to start, so I try to be the kind of guide to others that I wished I'd had along the way.
• Outside of code and data, you'll usually find me buried in a book. I'm a serial reader, mostly mysteries, thrillers, and romance novels, and I'm always up for a recommendation.

Get in Touch

Feel free to explore my repositories, and please don't be a stranger. I'd love to hear from you if you want to collaborate on a project, talk through an idea, or just swap notes on data science, machine learning, or AI. You can find me on LinkedIn or reach me directly at etsubwende@gmail.com.

Thanks so much for stopping by. I hope something here made you curious enough to look a little closer.

"If, at first, you do not succeed, call it version 1.0." (Khayri R.R. Woulfe)