Sergey Malchevskiy's Professional Page

About Me

I'm a result-focused professional with 15+ years in high-tech industry with expertise in machine learning and data analysis applying to finance, real estate, computer vision, and online products. More than 7 years I've been working in managing and leading roles for creating and implementing the digital products.

Since 2014 I've been researching and creating algorithmic trading strategies on cryptocurrency and stock markets. I began to carry out research as a self-started researcher. Also, I worked at a private algorithmic trading fund as a data scientist, and as a quantitative researcher at a company that develops an automated crypto trading platform ($20+ bln. monthly traded volume). Currently, we develop end-2-end algorithmic trading strategies.

When I began to share my ideas about the application of data analysis technologies with the community, I began to receive requests from companies to test their hypotheses. After that, I founded a company providing consulting services in this area.

Some of my works you can find on github, project and article sections on this website, or on my Medium's blog.

Latest Projects

Quantitative Stock Market Research

A lot of experiments were carried out to seek stable, scalable, and profitable algorithmic strategies. Different approaches are tested (pair trading, indicator systems, machine learning algorithms and so on). Some part of the results has positive return and alpha.

The used technologies: R, Python, Quantopian.

Mean Reversion Strategy

Volatility-based Switching Regime Strategy

Fit Realty

This project estimates the fair price of apartments of the real estate market in Saint Petersburg. Used machine learning algorithms allow to understanding what features have an impact on the price and what features haven't.

Also, it helps to search for undervalued objects to buying for housing or speculation.

Go to project

Noise Reduction Optimizer

Noise reduction optimization software for trading experts. Allows providing an analysis of time series in massive datasets on tick data (hundreds of millions of data points). The software estimates the results using statistical methods, generates the conclusions that help the expert to make a decision in trading operations.

Technological solution: the complex software with Walk-Forward analysis and Renko optimization inside it, global parameters searching using Bayesian optimization, high-performance Numba computations implemented on scalable multi-cores workflow, and so on.

QuizGeek

QuizGeek is the ultimate trivia app that will become your guilt-free addiction. The geeks behind QuizGeek have developed some sophisticated algorithms that learn the types of trivia questions you like/dislike so the game gets better and better the more you play. The game includes well-designed level progression, multiple styles of game-play, and a thought-provoking power-up strategy.

Go to project

CAPM Analysis

This service allows you to analyze the Russian stock market by CAPM for different time periods. A user receives a list of undervalued and overvalued assets, also calculated values of alpha, beta, Sharpe ratio and so on.

Go to project

Ident Smart Studio

The research project in pattern recognition field. ISS operates in three modes:

area-independent object analysis of ecosystem;
exterior analysis of biological species;
aerial photo counter system based on self-learning algorithms.

Go to project

ATM Network Analysis

The analytical project for Sberbank Data Science Journey based on ATM usage dataset. The are two ideas were proposed:

optimization the efficiency of ATM network usage based on a time series prediction of cash flow out value;
optimal reducing the number of ATMs in Moscow based on cluster analysis.

Find out more

Go to project

Live Picture

R&D project for Intel App Innovation Contest. We proposed and implemented additional dimension for traditional All-in-one PC. We added sensor for 3D-orientation based on Arduino. As a final product, we developed an application on Unity3D that demonstrates new possibilities.

Go to project

AlfaLike

AlfaLike is the social product for one of the largest commercial bank of Russia Alfa Bank. This product uses game mechanics to increase a loyal audience and involve it in banking products.

Read article

Work Experience

Founder - AI Works Lab (2020 - Present)

We are democratizing the adoption of AI technologies for business.

Drop a line if you have any ideas.

Co-Founder - Quant Epoch (2019 - Present)

We focus on creating a profitable, low-risk, and long-term stable trading mechanism. Our technology is a multi-layered optimization mechanism is used to ensemble a large number of low and medium correlated micro-strategies.

Technical Product Lead - Speech Technology Center (July 2019 - December 2019)

Leading the team (15+ experts) which is a part of the large R&D department in the AI company. The result of teamwork is an SDK for facial biometry.

developed a prototype of passive one-modal anti-spoofing system;
proposed and designed a hardware solution of passive anti-spoofing technology;
organized a process of large data collecting for ML models training using in-house service and crowdsourcing platforms (> 1 TB);
leading the research and software development teams according to the product roadmap;
controlling the roadmap changes and risks;
communicating and supporting external product teams and clients about using our technology.

Previous experience since 2007

Data Science Expert - Conjecture corp. (2015 - 2018)

Participation in the development of online products in educational and entertainment areas.

implemented a data-driven approach to decision-making process;
coordinated a process for data collection and attribution;
formulated and tested product hypotheses to get improved gaming and marketing indicators;
carried out a lot of ad-hoc analyses;
developed a dashboard system for monitoring and getting actual information about product (technical performance, user information, feature conversions, and so on).

Digital Director - Global Point Agency (2014 - 2015)

Managing a business unit in GPA (major advertising agency, head office in Saint Petersburg). We provided digital marketing consulting in 360 degrees.

achieved the positive profit of the unit in the crisis period;
formed the department consisted of groups of account managers, producers, software developers, and designers;
implemented a large number of projects for international and all-Russian companies (finance, automotive, tobacco, etc.);
won the awards in marketing and advertising.

Head of Software Development - Ascreen IM (2012 - 2014)

Building a technical department that consists of software development and QA groups in the interactive media company. We implemented R&D projects and completed commercial products for our clients. The basic stack of technologies: real-time graphics, gesture based systems, AR applications, multi-touch applications for interactive tables, etc. Basically, our clients were large industrial companies, advertising agencies, and museums.

Software Engineer -> Team Lead - Center of IT, NSII (2008 - 2012)

Development, implementation, and support of IT system in an educational organization. We implemented ERP Galaktika as a basic educational business platform and integrated it with specialized software platforms (1C, MARC bibliographic system). Some parts of our architecture with in-house developed software were based on SOA and ASP.NET services.

Software Engineer Intern - JSC Norilsk Nickel (2007 - 2008)

Created an information system for storing graphical objects in hierarchical representation for mine department. Also, I participated in migration process from MSSQL to Oracle database as basic DBMS of ERP platform. I wrote a few scripts with business logic on PL/SQL and VBA.