dvukolov Goto Github PK

followers: 5.0 following: 1.0 repos: 24.0 gists: 1.0

Name: Dmitry Vukolov

Type: User

Location: Belgrade, Serbia

Blog: linkedin.com/in/dvukolov

Hi there 👋

I'm a data analyst passionate about coding in Python, R, and Go.

Some of my projects include:

Calibrated Uncertainty (NumPy, SciPy, scikit-learn, PyMC3, JAX-based NumPyro)

Conducted in-depth analysis of an uncertainty calibration algorithm for Bayesian neural networks.
Identified the advantages of the approach and its modes of failure.
Received the highest grade and an invitation to continue research at the Harvard Data to Actionable Knowledge lab.

Galaxy Measurements (TensorFlow, pandas, Streamlit, Docker, Heroku)

Used TensorFlow to estimate the shape and brightness of simulated galaxies.
Responsible for data generation, exploratory data analysis web app, neural architecture search, and denoising pipelines.
Got the top grade and an opportunity to continue research.

NBA Conference Advantage (R, tidyverse, Scrapy, LaTeX)

Performed statistical modeling of potential bias in the NBA that grants teams in one conference an easier path to success due to the differences in travel and schedule.
Wrote web scrapers, feature engineering, and most of the analysis code in R.
Built linear regression models, ran diagnostics, authored around 70% of the report.

Meteorological Observatory (Python, TCP sockets, regex, pytest)

Implemented in Python streaming data collection from weather instruments.
The code has a suite of unit tests and is deployed at a university meteorological station.
Serves as the basis for experimental studies of turbulence by five scientific institutions.

Open Source Projects (Python, R, C/C++)

Contributed with bug fixes and documentation updates to open source projects such as Apache Arrow, scikit-learn, ThunderSVM (GPU accelerated SVM), Picasso (sparse regression algorithm), Gap Statistic (clustering metric).

Dmitry Vukolov's Projects

arrow

Apache Arrow is a cross-language development platform for in-memory data. It specifies a standardized language-independent columnar memory format for flat and hierarchical data, organized for efficient analytic operations on modern hardware. It also provides computational libraries and zero-copy streaming messaging and interprocess communication. Languages currently supported include C, C++, Java, JavaScript, Python, and Ruby.

calibrated-uncertainty

Analysis of the uncertainty calibration algorithm for Bayesian neural networks

calibration-research

Spring 2020 Research into Uncertainty Metrics and Calibration

decimal

A high-performance, arbitrary-precision, floating-point decimal library.

dvukolov

fsnotify

Cross-platform file system notifications for Go.

galsim

The modular galaxy image simulation toolkit. Documentation:

gap_statistic

Dynamically get the suggested clusters in the data for unsupervised learning.

gotop

A terminal based graphical activity monitor inspired by gtop and vtop

jupyter-black

Black formatter for Jupyter Notebook

kedro

A Python library for building robust production-ready data and analytics pipelines

keras-tuner

Hyperparameter tuning for humans

measure-galaxies

Estimating the shape and flux of galaxies with neural networks

minimc

Just a little MCMC

nba-conference

Statistical modeling of NBA conference bias due to differences in travel and schedule

nefit-easy-http-server

HTTP server to access Nefit/Bosch XMPP backend over HTTP

picasso

Penalized Sparse Learning Solver - Unleash the Power of Nonconvex Penalty

s3fs

S3 Filesystem

scikit-learn

scikit-learn: machine learning in Python

scrapy-mongodb

MongoDB pipeline for Scrapy. This module supports both MongoDB in standalone setups and replica sets. scrapy-mongodb will insert the items to MongoDB as soon as your spider finds data to extract.

streisand

Streisand sets up a new server running your choice of WireGuard, OpenConnect, OpenSSH, OpenVPN, Shadowsocks, sslh, Stunnel, and a Tor bridge. It also generates custom instructions for all of these services. At the end of the run you are given an HTML file with instructions that can be shared with friends, family members, and fellow activists.