Michael Savin

Data / ML Portfolio

Built a finance-focused risk engine around earnings events.

I build data pipelines, engineer predictive features, validate signals out-of-sample, and turn the result into usable products. My flagship project is Breakwater - an earnings-period risk dashboard designed to surface concentrated tail-risk regimes.

Open Breakwater Dashboard GitHub LinkedIn
Python
Pandas
Time-Series Features
ML Validation
Streamlit
Finance / Risk
Breakwater Risk Snapshot Live demo

The Breakwater Pipeline

From Data to Earnings Risk Intelligence

Breakwater processes price, earnings, EPS, and sector data into explainable risk scores, alerts, dashboards, and per-stock reports.

Technical depth

Multi-stage Python pipeline, time-series feature engineering, event-based labeling, validation logic, and score design grounded in risk concentration.

Clear signal quality

Focused on finding regimes where large post-earnings moves become much more likely, instead of pretending direction can always be forecasted.


1

Collect & Normalize Market Data

Fetches and standardizes data for current S&P 500 stocks.

  • Stock prices
  • Earnings dates
  • EPS actuals & estimates
  • Sector information
2

Engineer Market Features

Transforms raw time-series data into event-aware market signals such as:

  • 3-day Reaction
  • Stock Drift
  • Stock Volatility
  • Sector Momentum
3

Build Risk Scoring Features

Combines higher-order signals into an explainable earnings risk engine.

  • Earnings explosiveness
  • Momentum pressure regime
  • Volatility expansion
  • Composite risk score
4

Deliver Insights

Serves outputs through a deployed product interface.

  • Streamlit dashboard
  • Risk alerts
  • Stock drill-downs
  • PDF reports
500 Stocks covered
25+ Years of data
50+ Engineered features
Live Dashboard deployment

Other Projects

Breakwater is the centerpiece. Here are other projects of mine.

Mortgage Approval Classification Project

ML / Classification

End-to-end binary classification workflow for loan approval prediction, including preprocessing, statistical exploration, and model evaluation.

GitHub

Neural German Translation with Transformers Project

Deep Learning

Coursework-driven neural network project showing practical familiarity with modern ML tooling, model training flow, and evaluation logic.

GitHub

Tel Aviv University Research - Music Data / N-gram Analysis

Research / Data Analysis

Structured symbolic music corpora into analyzable data pipelines, extracted root progressions, and built composer-level comparison outputs from large annotation sets.

GitHub

Skills

Programming

  • Python
  • Pandas / NumPy
  • Modular pipeline design
  • Data wrangling

Machine Learning

  • Classification workflows
  • Feature engineering
  • Evaluation / AUC / correlation
  • Out-of-sample validation

Product / Analytics

  • Risk-oriented scoring systems
  • Dashboard thinking
  • Interpretability
  • Report-ready outputs

Tools

  • Streamlit
  • Git / GitHub
  • Tableau
  • SQL basics

Background

Education

B.Sc. in Computer Science & Musicology, Tel Aviv University.

  • Data Analysis
  • Machine Learning
  • Deep Learning for Data Science
  • Databases / Information Systems

Additional Experience

  • Research assistant work in music-data analysis
  • 9+ years of tutoring in math, physics, and CS.
  • Strong communication, technical explanation, and structured thinking.

Contact

If you'd like to discuss Breakwater, my project work, or data / ML roles, reach out here.

Michael Savin

Data / ML candidate focused on applied analytics, predictive systems, and finance-oriented tooling.

Email GitHub LinkedIn Breakwater