Skip to content
View sergey-kasatov's full-sized avatar
1 follower · 26 following

Block or report sergey-kasatov

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
sergey-kasatov/README.md

Sergey Kasatov - Data Analyst

Location: Cologne, Germany
Stack: SQL • Python (pandas, scikit-learn) • TensorFlow / Keras • Tableau • Streamlit • MLflow
Domain: Manufacturing & Quality analytics
Open to: Data Analyst / BI Analyst / Operations Analyst roles in Germany from September 2026

Links

Featured work

  • FreshMart Demand Planner - Time Series Forecasting App (Python + MLflow + Streamlit)
    Built an end-to-end sales forecasting pipeline with EDA, feature engineering, statistical models, machine learning models, HyperOpt tuning, MLflow experiment tracking, and a deployed Streamlit planning app.
    Repo: time-series-forecasting-project · Live App: FreshMart Demand Planner

  • Disaster Tweets - NLP Classification (Python / scikit-learn / PyTorch / DistilBERT)
    Binary classifier that distinguishes real-disaster tweets from figurative use. Shipped an interpretable TF-IDF + tuned Logistic Regression baseline, benchmarked against a from-scratch CNN and fine-tuned DistilBERT; results confirmed on the Kaggle public leaderboard.
    Repo: nlp-disaster-tweets-classification

  • Phoenix Retail - KPI Dashboard (Sales, Profit & Discount)
    Built a KPI dashboard with profit/discount analysis and clear business takeaways.
    Repo: phoenix-retail-kpi-dashboard · Tableau: Retail KPI Dashboard - Main Dashboard

  • TravelTide - Customer Segmentation & Retention (Python + SQL + ML)
    Built a customer retention segmentation framework using SQL, Python, rule-based perk assignment, and ML benchmarking, then translated the results into business-ready targeting recommendations.
    Repo: traveltide-segmentation-retention

  • CIFAR-10 ResNet50 Transfer Learning Experiments (Python / TensorFlow / Keras)
    Systematic experiments on dataset size, augmentation, LR scheduling, and partial unfreezing. Best result: 82.7% test accuracy. Includes Grad-CAM visualization.
    Repo: cifar10-resnet50-transfer-learning-experiments

  • Predictive Maintenance - Maintenance Required Prediction (Python / scikit-learn)
    Baseline model + metrics to predict maintenance need from usage/condition signals.
    Repo: predictive-maintenance-ml · Dataset: Kaggle - Logistics Vehicle Maintenance History

Pinned Loading

  1. portfolio portfolio Public

    Portfolio hub: SQL, Python, BI dashboards, ML, forecasting and Streamlit projects

  2. time-series-forecasting-project time-series-forecasting-project Public

    End-to-end time series forecasting project for grocery sales, including EDA, feature engineering, statistical and ML models, HyperOpt tuning, MLflow tracking, and a Streamlit planning app.

    Jupyter Notebook

  3. traveltide-segmentation-retention traveltide-segmentation-retention Public

    Customer segmentation & retention case study using SQL, Python, rule-based perk assignment, and ML clustering benchmark.

    Jupyter Notebook

  4. cifar10-resnet50-transfer-learning-experiments cifar10-resnet50-transfer-learning-experiments Public

    ResNet50 transfer learning on CIFAR-10: from a 66% baseline to 82.7% through systematic, controlled improvements.

    Jupyter Notebook 1

  5. predictive-maintenance-ml predictive-maintenance-ml Public

    Predictive maintenance ML case study (Python / scikit-learn)

    Jupyter Notebook