Portrait of Jemuel Oppong

Jemuel Oppong

Data Science & Statistics @ University of Michigan

About me

I’m a Data Science student with a growing focus on machine learning research and its potential to advance both AI capabilities and real-world applications. I’m particularly interested in developing systems that can improve upon existing technologies, enhance the way people work, and address complex challenges across industries. My long-term goal is to contribute to ML research that makes AI more capable, reliable, and impactful, while helping shape how the technology evolves and benefits society.

Projects

In progress

Generative 3D Cities

MDSTSept 2026 – Present
  • Engineered a procedural city generator in Python (NumPy, SciPy) producing 128x128 four-channel grids (terrain height, water mask, building height, land use), exported via Blender (bpy) as colored, 3D-printable .blend/.stl meshes.
  • Trained a PyTorch multi-head model on 15,000 synthetic prompts with frozen sentence-transformer embeddings to predict 10 generator parameters from natural language, evaluated on held-out and hand-written prompts and a road-continuity metric.
  • Shipped a Streamlit app with prompt and manual modes, 3D preview, and PNG/STL/GLB export; extending it into a two-stage conditional model where terrain output conditions a building model to respect water and slope constraints.
PythonPyTorchNumPySciPySentence-TransformersBlender (bpy)Streamlit
Click to check status

Macomb County IDEP Data System

Michigan Data ConsultingSept 2026 – Present
  • Engineering a prototype map and dashboard for a water-quality data system built on E. coli and microbial source tracking data, for the Macomb County Public Works Office, serving 800K+ residents and businesses.
  • Designed a relational data model and standardized location-coding schema, cleaning and deduplicating five years of fragmented data records into a reproducible pipeline that supports trend analysis and future predictive modeling.
SQLETLData Cleaning & ModelingStatistical Data AnalysisRelational SchemaGISInteractive Dashboard
Music by 9JackJack8 from Pixabay