Mariia Matskevichus
Data Scientist - Grid Dynamics
Mariia is a data scientist at Grid Dynamics, Saint-Petersburg, Russia. She is also a final year master at ITMO University at Big Data and Data analysis program.
Mariia has experience and knowledge in visual search, NLP, anomaly detection and time series forecasting with Python. She takes participation not only in enterprise projects but in R&D, writing academic papers and giving lectures at local universities.
In her free time, she enjoys running, traveling and filter coffee.
PyCon Balkan 2019 Talks
TLK PyCon Talk
How to train a neural network to classify screws by photo
Imagine, thousands of bolt and screw positions in the catalog, bulky filters, unclear product descriptions… all this is a headache for the retailer and its customers.
How to improve the customer journey through e-catalog and bring them to purchase? Let’s make a visual search app to find exactly the same screw or bolt by a photo only! And Python gives us a full set of instruments for doing that.
The main difficulty of the solution lies in the scalability of the screw number that the app can recognize. After the presentation, you will get the full pipeline of building the visual search system from a scratch with Python.