PORTFOLIO PROJECTS

Build Real Things

Three production-grade projects that go in your GitHub and your university applications.

01
PROJECT 01 — GENOME CLASSIFICATION

Genome Species Classifier

Train a machine learning model on real DNA sequences from NCBI to classify which organism a sequence belongs to. Deploy it as an interactive Streamlit web app that anyone can use.

scikit-learnBiopythonPandasStreamlitNCBI API
  • 1Download genome sequences from NCBI GenBank (5 organism classes)
  • 2Engineer k-mer frequency feature vectors (k=3, k=4)
  • 3Train Random Forest + XGBoost comparison with cross-validation
  • 4Evaluate with ROC curves and confusion matrices
  • 5Deploy as Streamlit app — paste any DNA, get prediction instantly
Get Project Materials → GitHub Template
ATGCGATCGATCGGCTATCG... ATG:0.08 GCG:0.05 TAT:0.03 CGA:0.07 GAT:0.04 GGC:0.02 RANDOM FOREST (200 trees) Tree1 Tree2 Tree3 ... E. coli confidence: 94.2%
CONTACT MAP PREDICTION Predicted 3D contacts → fold structure
02
PROJECT 02 — PROTEIN STRUCTURE

Protein Contact Map Predictor

Using multiple sequence alignment (MSA) coevolution signals, predict which amino acids are in physical contact in 3D space — the same technique used before AlphaFold2 was released.

PyTorch ESM-2 Biopython py3Dmol
  • 1Fetch protein sequence from UniProt
  • 2Build MSA using BLAST homolog search
  • 3Compute mutual information coevolution matrix
  • 4Compare predictions to PDB ground truth
  • 5Visualize contacts on 3D structure with py3Dmol
Get Project Materials →
03
PROJECT 03 — MEDICAL AI

Medical Anomaly Detector

Build an autoencoder trained on healthy chest X-rays that detects anomalies. Add Grad-CAM heatmaps showing exactly where the AI looks, and deploy it as a FastAPI REST service.

MONAI PyTorch Grad-CAM FastAPI Docker
  • 1Download NIH Chest X-ray dataset (normal subset)
  • 2Train ResNet18 autoencoder on normal images only
  • 3Flag high reconstruction-error images as anomalies
  • 4Generate Grad-CAM heatmap overlays
  • 5Package as Docker container and deploy FastAPI endpoint
Get Project Materials →
ANOMALY DETECTION PIPELINE X-RAY INPUT ENCODE z DECODE RECON OUTPUT MSE = 0.047 ⚠ ANOMALY DETECTED

Start Building Today

Get all project templates, datasets, and step-by-step guides when you enroll.

Enroll Free →