Skip to content
View Meta1rus's full-sized avatar

Block or report Meta1rus

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.

Content in all repositories owned by your account will be closed.
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
Meta1rus/README.md

Filipp Podolskii — Data Analyst, BI

Website fpodolskii.com Perm, Russia Open to remote work

👋 About Me

  • Data analyst focused on BI: SQL, Power BI, Excel / Power Query, 1C, Python
  • Master's student in Applied Mathematics and Informatics at Perm State University
  • Looking for a full-time remote BI Analyst position

🎓 Education

  • Master's degree (in progress), Perm State University (2025 — present)
    • Field of study: Applied Mathematics and Informatics
  • Bachelor's degree, Perm State University & Moscow Institute of Physics and Technology (MIPT) (2021 — 2025)
    • Fields of study: Applied Mathematics and Informatics; Artificial Intelligence and Big Data

🛠 Tech Stack

BI & Analytics

SQL Power BI Excel Power Query 1C

Programming & Data

Python pandas NumPy Jupyter Matplotlib Plotly

Machine Learning

scikit-learn CatBoost PyTorch Hugging Face

Databases & Data Engineering

PostgreSQL SQLite DuckDB Docker Git n8n Scrapy Playwright

📊 Projects

Model comparison on the test set
Purchase prediction for online store sessions
EDA → 7 models → SHAP → dashboard in Docker
CatBoost: ROC-AUC 0.929 on the test set 		Repository popularity by GitHub topic
Web data collection and an n8n pipeline
REST API, Scrapy, Playwright, data quality checks, orchestration
Data on 592 repositories with zero rejected records
NDCG@10 on the test set
Two-stage recommender system
iALS and Mult-VAE candidates + CatBoostRanker on KION data
NDCG@10 14% higher than the best single model 		Correlation heatmap from the bot's reply
Telegram bot for exploratory analysis of CSV files
Summary, charts and an Excel report for an uploaded table
Covered by 26 offline tests
Signal forecast with a confidence band
Kalman filter signal forecasting and 16 time series labs
AR(2), SARIMA, Prophet, DTW, ROCKET, echo state networks
Useful forecast horizon 7 s vs 1.2 s for the naive forecast 		t-SNE and UMAP, colour shows diagnosis
Unsupervised clustering: 24 pipelines
UMAP, t-SNE, PCA, Isomap, stability analysis
94% agreement between clusters and diagnosis
t-SNE of the network's hidden states
Russian sentiment analysis and 10 PyTorch practices
BiLSTM + rubert-tiny2, CNN, transfer learning, VAE, transformers
macro-F1 0.751 on the test set 		All repositories →
every README states the result; data is downloaded by scripts; project READMEs are in Russian

Popular repositories Loading

  1. web-data-extraction web-data-extraction Public

    Сбор данных из интернета: REST API, BeautifulSoup, Scrapy, Playwright, контроль качества и гибридный пайплайн GitHub Topics + API с оркестрацией в n8n

    Jupyter Notebook

  2. python-data-analysis python-data-analysis Public

    Telegram-бот для разведочного анализа CSV (pandas, matplotlib, Excel-отчёт) и 13 задач на Python с автотестами

    Python

  3. clustering-dimensionality-reduction clustering-dimensionality-reduction Public

    Кластеризация опухолей без учителя: 24 пайплайна понижения размерности и кластеризации (UMAP, t-SNE, PCA, Isomap) с анализом устойчивости — совпадение с диагнозом 94 %

    Jupyter Notebook

  4. eda-classification eda-classification Public

    Прогноз покупки по сессии интернет-магазина (UCI 468): EDA по CRISP-DM, CatBoost (ROC-AUC 0,929), SHAP и дашборд в Docker

    Jupyter Notebook

  5. neural-networks neural-networks Public

    Тональность русскоязычных текстов: BiLSTM на подсловах rubert-tiny2 (macro-F1 0,751) и 10 практик по нейросетям на PyTorch — CNN, transfer learning, VAE, трансформеры

    Jupyter Notebook

  6. time-series-regression time-series-regression Public

    Временные ряды: прогноз сигнала фильтром Калмана с моделью AR(2) (полезный горизонт 7 с) и 16 работ — SARIMA, Prophet, DTW, ROCKET, эхо-сети, звук

    Jupyter Notebook