My name is Stanislav Khrapov. I am a full stack Machine Learning engineer with many years of experience building data science applications starting from data ingestion, modelling, and finishing with cloud deployment and monitoring services. I am passionate about data, automation, code quality, visualisation, and communicating with both technical and non-technical stakeholders.
Currently, I work as a lead data scientist at a small consulting company M2hycon. In this role I have led a cross-functional team of data scientists and data engineers designing and implementing a dynamic pricing application based on Bayesian methods. I have also taken over a project with a lot of legacy code and replaced an existing failure detection model that was using a neural network with a much simpler, faster, and more accurate classification algorithm.
In the past I was a senior data scientist at the FinTech startup Chintai based in Frankfurt am Main, Germany. My main project is the development of trade surveillance system based on unsupervised time series classification machine learning models. In the absence of comprehensive training data we have built realistic exchange market simulation with heterogeneous traders. One part of it is the fast Python-based order book matching engine (OrderBookMatchingEngine). To tie all pieces together we have designed a fully automated and reproducible pipeline to simulate market data, train and evaluate ML models catching illegal trading behaviour. As in any startup I also did a little bit of everything IT-related. This includes building GitHub actions based CI/CD pipelines to release Node.js web applications interacting with the blockchain, check code quality, unit and integration tests, search for code vulnerabilities, deploy to Kubernetes cluster running in the cloud. I have also organized processes across the company to streamline and speed up development and release activities starting from a PR and finishing with deployment to the production environment.
I started my career in data science at DB Schenker also based in Frankfurt. There I designed new time series models for forecasting of market freight prices and volumes, company internal financial indicators (EBIT, revenue, receivables, payables, etc.). My responsibilities also included writing end-to-end data ingestion, processing, forecasting, and delivery software mainly in Python and using such tools as web scraping, SQL, pandas, scikit, GitLab, Docker, AWS, Azure, Airflow, etc. I performed research on model comparison in terms of forecasting performance. On top of that I love doing sophisticated visualizations (Seaborn, Dash) for presentation to internal business clients. Finally, I regularly worked as an instructor for the internal AI Training Workshop.
Even before that, as I was working as Assistant Professor of Finance at the New Economic School in Moscow, Russia. My area of specialization was financial econometrics, option pricing, volatility modeling. During graduate education and work experience that includes both academic and industry positions I wrote research papers individually and in collaboration.
On the personal level I run and swim a lot, bike occasionally (no triathlon, please!). You can join me in this passion on Strava.