Importants
"Cyfrovoi proriv 2022" hackathon 3rd place winner (team: I PRO) - link
Work done:
- Fixed data disbalance
- Composed new deep model architecture based on adding heads to ResNet34 model
- Trained model to perform inference
Technologies used:
- PyTorch, TorchVision
- ResNet34
- FastAPI
Project for Rosselhoznadzor
Work done:
- Maintained communication with the customer, drawing up the necessary data requirements and explaining the results obtained by the model
- Visualised separation of classes on PCA
- Enabled autonomous work of model on mobile device
Technologies used:
- Dart
- Matplotlib
- sklearn
Project for EESTECH CHALLENGE
Work done:
- Used OpenAI API and TF-IDF to provide quality answers to predefined knowledge database
- Expanding database in runtime
Techonologies used:
- OpenAI API
- TF-IDF
- aiogram