BubCNN is a bubble detection system which employs a pretrained Faster RCNN module to locate bubbles and a pretrained shape regression CNN module to approximate the bubble shape by an ellipse.

Assembled Phi-X174 genome using Overlap Graph, Kmer Composition and De-Bruijn Graph.

Full-stack pipeline to detect U.S. housing market bubbles and forecast price trends. Merges 6+ macroeconomic datasets in Snowflake to compute risk scores and price predictions using walk-forward ML models. Deployed with Streamlit for interactive insights.

A Wolfram Mathematica image processing code for bubble flow images in X-ray. Codeveloped with https://github.com/Mihails-Birjukovs

End-to-End Python implementation of LPPLS (Log-Periodic Power Law Singularity) framework for detecting financial bubbles and critical transitions. Features Filimonov-Sornette calibration, Lagrange regularization, Lomb-Scargle spectral validation, and Monte Carlo significance testing. Complete computational replication of Hosseinzadeh (2025).