Ivan Karlov
#
karlovivan6799@gmail.com |
+7 (996) 205-05-03 |
}
@IvanK2003 |
§
DaEtoJostka
Summary
Data Scientist with over 2,5 years of experience in machine learning and data analysis. Seeking a chal-
lenging position to apply my expertise in data science to help improve business outcomes.
Work Experience
Data Scientist at IndorSoft February 2022 – September 2024
• Developed a CatBoost ensemble to redirect support requests to appropriate FAQ sections, reducing
team workload by 15%.
• Developed an add-on for IndorTrafficPlan to recognize and transfer road signs from video to road plan
(demo link):
– Trained YOLO for detecting road signs (F1 = 0.96, mAP@0.5–0.95 = 0.89).
– Developed classifiers based on EfficientNet V2 for sign groups (accuracy ≥ 0.98).
– Created a hybrid model with MobileNet V3 for tabular and visual data, reducing inference time by
3x without accuracy loss.
– Implemented an LSTM model to improve tracking, eliminating ID Switches.
– Generated synthetic data to expand dataset by 4x, improving detection by 8% (Recall), precision by
12% (IoU), and classification by 10% (F1).
– Assisted in developing a triangulation method for accurate distance measurement to signs (±0.15 m).
– Developed scripts for classifier and detector interaction.
– Corrected data annotation using CVAT and managed datasets with FiftyOne.
Data Scientist Intern at IndorSoft January 2022 – February 2022
• Wrote efficient SQL queries for data extraction and analysis.
• Refactored existing code.
• Searched, analyzed, and prepared datasets.
• Presented results using Gradio.
• Developed an image clusterizer using ViT and K-means, achieving Silhouette Score = 0.91 after PCA.
Improved YOLO classifier by 12% (F1) with new data splitting.
Skills
Git, Python, Plotly, PyTorch, NumPy, Docker, Linux, PostgreSQL, CatBoost, TensorFlow, Keras, OOP,
SQL, Machine Learning, Pandas, Data Science, Big Data, Scikit-learn, Computer Vision, Data Mining,
NLP
