Artem Budishchev verfügbar

Artem Budishchev

Machine learning and artificial intelligence expert with 5 years of industry experience in FinTech

verfügbar
Profilbild von Artem Budishchev Machine learning and artificial intelligence expert with 5 years of industry experience in FinTech aus Hamburg
  • 20149 Hamburg Freelancer in
  • Abschluss: Vrije Universiteit Amsterdam, Amsterdam, the Netherlands
  • Stunden-/Tagessatz:
  • Sprachkenntnisse: deutsch (Grundkenntnisse) | englisch (verhandlungssicher) | russisch (Muttersprache)
  • Letztes Update: 20.05.2020
SCHLAGWORTE
PROFILBILD
Profilbild von Artem Budishchev Machine learning and artificial intelligence expert with 5 years of industry experience in FinTech aus Hamburg
SKILLS
Machine learning: supervised learning, unsupervised learning, reinforcement learning, deep learning,  ensemble learning, semi-supervised learning, transfer learning.
Programming languages ​​and frameworks: Python, R, SQL, NoSQL, bash; Python frameworks: numpy, pandas, scikit-learn, TensorFlow, scipy, matplotlib, bokeh, XGBoost, LightGBM, CatBoost, Jupyter Notebooks R frameworks: ggplot2, data.table, knitr, parallel, dplyr, shiny and other tidyverse packages. 
Big data and machine learning tools: Spark/PySpark, Databricks,  
Databases:  Postgresql, MS SQL Server, EXASOL, mongodb, redis.
Cloud platforms: Amazon AWS, Microsoft Azure.
Container orchestration tools:  Docker, Kubernetes, jenkins.
Other: end-to-end delivery of machine learning models from ideation to production stage, including research and prototyping, data pipelines and model building, deployment of the finalized models to production (MLOps).
PROJEKTHISTORIE
  • 12/2019 - bis jetzt

    • MASH Luxembourg
    • 50-250 Mitarbeiter
    • Banken und Finanzdienstleistungen
  • Principal Data Scientist

  • 01/2018 - 12/2019

    • collectAI GmbH (subsidiary of Otto Group)
  • Senior Data Scientist
  • ? Delivered end-to-end data science projects, including core business ML solutions.
    ? Researched and implemented state-of-the-art methods for the next generation of AI solution.
    This includes deep reinforcement learning, deep learning (LSTM, ordinary feedforward
    networks), and gradient boosting. The deep reinforcement learning algorithm
    allowed one of our clients to boost collections by about 14%. This resulted in monetary
    gains of nearly 65000EUR p.a.
    ? Developed a contextual bandit algorithm for clients with smaller amounts of data. Average
    uplift - 2%.
    ? Developed a message classi?cation algorithm that allowed to reduce operational workload
    for the clients.
    ? Led a team of 5.
    ? Created a roadmap for the data science projects.
    ? Devised data collection/data quality control strategy and collaborated with multiple
    teams to re?ne data collection practices.
    ? Communication of the needs and results of the data science team to internal, as well as
    external (Otto Group board members) stakeholders.
    ? Building solutions to benchmark/backtest the performance of the ML algorithms.




    ? Building solutions for internal (operations, product managers, etc.) and external (customers)
    stakeholders for reporting and performance (KPI) monitoring of the customers'
    portfolios.
    ? Supporting the sales team with ad hoc analyses for customer retention and acquisition.
    ? Participating in the organization committee for the HamburgAI events.
    ? Technology stack: Python (and ecosystem, e.g. numpy, tensor ow, keras, sanic as async
    microframework) to productionize the model, R (and ecosystem, e.g. ggplot2, data.table,
    etc.) for PoC, exploratory analysis + reports. Kubernetes in AWS for deployment to
    production.

  • 04/2016 - 12/2017

    • Kreditech Holding SSL GmbH
  • Data scientist
  • ? Credit risk predictive modeling using gradient boosting. By applying a multi-task technique,
    I was able to integrate a relatively new data source to the main dataset and reduce
    default rate on backtest by more than 1%.
    ? Developed a programmatic solution for optimal bidding strategy in search engine advertisement
    campaigns. Multiple experiments showed nearly 20% cost reduction on Google
    AdWords.
    ? Predictive modeling using logistic regression/ensembles of decision trees for multi-channel
    online attribution problem.
    ? Predictive modeling of o?er dropouts.
    ? Collaborated with multiple teams to improve data quality of the customer journey dataset.
    ? Designed and implemented a data processing pipeline for customer journey dataset tailored
    to business needs.
    ? Designed and implemented a data processing pipeline for risk modeling.
    ? Supported various teams on decisions for test & learn campaigns.
    ? Analyzed the performance of a?liates and provided insights into improvement of the
    a?liate bonus schema.
    ? SEO improvement by providing insights through Google search terms analysis of competitor
    rankings.
    ? Development of software packages in R and Python.
    ? Building simple dashboards using bokeh (Python) and knitr (R).
    ? Built Jenkins/Docker infrastructure to support various tasks within data science department.

    ? Developed a monitoring software to identify problems in the conversion funnel using various
    KPIs, such as conversions, costs, CPA etc., that automatically noti?es the stakeholders
    of any abnormalities in the KPIs.
    ? Participated in hiring process for new data scientists.
    ? Led a sub-team of two junior data scientists.

  • 08/2015 - 04/2016

    • Vacansoleil
  • Software developer
  • ? Back-end and front-end development in Python, JavaScript, SQL, HTML/SASS using
    Django, Djangocms, and Postgresql, as well as jQuery and Bootstrap frameworks in a
    Scrum environment.

  • 02/2011 - 05/2015

    • VU University
  • PhD-student
  • ? Conducted predictive and explanatory modeling of climate data using machine learning
    algorithms such as arti?cial neural networks, random forests, gradient boosting, support
    vector machines and extreme learning machines using Python, R, MATLAB, Fast Arti?-
    cial Neural Networks (FANN) library and Vowpal Wabbit.
    ? Worked with large datasets of high-frequency data. Applied data wrangling/cleaning
    and outlier detection on dense data. Used hypothesis testing for feature selection. Successfully
    developed a cross-validation-based framework for predicting climate data, which
    signi?cantly improved prediction error estimates.
    ? Independently initiated research projects, which led to publications/chapters in the PhD
    thesis.

  • 09/2007 - 01/2011

    • Yu. G. Shafer Institute of Cosmophysical Research and Aeronomy
  • Junior Researcher: Atmospheric physics
  • ? Worked with large datasets of geospatial (satellite) data. Successfully implemented a fully
    automated data storage/data processing/data visualization system (bash + Python).
    ? Analyzed geospatial datasets using ArcGIS and QGIS.
    ? Co-taught `Relational database management systems (RDBMS)' course for undergraduate
    students.

ZEITLICHE UND RÄUMLICHE VERFÜGBARKEIT
In general, I'm flexible on the location, but would prefer remote work with occasional on-site visits.
KONTAKTANFRAGE VERSENDEN

Nachricht:

Absenderdaten: