Ayush Aary

Hi! I’m Ayush Aary, a Civil Engineering undergraduate at IIT Roorkee exploring Machine Learning, and Software Engineering.

I’m currently building strong fundamentals in Python and C++ while learning DSA and applied ML. I enjoy working on problem-solving tasks and data-driven projects, and I like understanding both the technical and business side of problems.

Skills:

• Computer languages: Python, C++
• Software Packages: Pandas, NumPy, Seaborn, Matplotlib
• Currently learning: Machine Learning, DSA
• Tools: Git/GitHub, Google Colab

Projects:

• Satellite Imagery–Based Property Valuation (Multimodal ML)
  Built a multimodal price prediction system combining structured housing data with satellite imagery. Extracted visual embeddings using pretrained ResNet18, reduced them via PCA, and fused with engineered tabular and spatial features in an XGBoost regressor. Performed extensive EDA and feature engineering (log-scaled sqft, house age, zipcode offset, structural density). The multimodal model achieved consistent improvement over tabular-only baseline by capturing neighbourhood-level context from satellite images.

• Geospatial Landslide Risk Mapping (Safe Hills Uttarakhand)
  Built an AI-powered landslide susceptibility system using DEM, slope, NDVI (Sentinel-2), rainfall, geology, and drainage data. Trained a Random Forest model and combined it with a Weighted Landslide Susceptibility Index (LSI) to map high-risk zones across Uttarakhand. Developed an interactive Google Earth Engine dashboard with layer inspection and next-day rainfall forecasting. Validated the model on the Chamoli landslide case, correctly identifying it as a high-risk region

  What I bring to a hackathon:

• Strong analytical thinking + ML fundamentals

• Experience with geospatial and multimodal data

• Comfortable turning messy datasets into insights

• Can contribute to both technical development and presentation/storytelling

• Fast learner and collaborative team member