
Senior Data Scientist at Experian · Sofia, Bulgaria
Discover more about me
I’m a Senior Data Scientist in Experian’s Proactive Analytics team in Sofia. I lead the Bulgaria-based team members and build tools that help analysts explain trends in U.S. consumer credit. My reporting platform connects the figures in each report to the calculations behind them.
At Nanyang Technological University in Singapore, I studied computer science, specialising in AI and data science with a minor in mathematics. For my final year project, I built a system that reads handwritten STEM exams, from page layout to mathematical notation and grading.
Experian recognised my reporting platform in the Individual Contribution category.
Experian
My handwritten exam platform placed in the top 10% of Computer Science submissions at the Global Undergraduate Awards.
NTU College of Computing and Data Science. Final-semester GPA of 5.00/5.00.
Google IT Automation with Python, Google Business Intelligence and Google Project Management
Bulgarian (native) and English (full professional proficiency)
View my education
Artificial Intelligence and Data Science, Minor in Mathematics
Completed NTU’s Accelerated Bachelor programme in 3.5 years with Highest Distinction and a cumulative GPA of 4.59/5.00.
Full-Stack Web Development with Python and Django
Completed an intensive and project-oriented web development program focused on building real-world Python and Django web applications, including RESTful APIs, database management, and frontend integration.
Mathematics, Informatics, and Programming
Specialized in mathematics and informatics under the Bulgarian curriculum. Participated in competitive programming and mathematics contests, and served in the student council, developing both technical and leadership skills.
View my professional experience
Experian
I lead the Bulgaria-based members of Proactive Analytics at Experian, delivering credit analytics and reporting for the U.S. market.
Explore my projects
A pre-release typing platform built around Bulgarian literature and historical figures.
A desktop tool that reads Bulgarian invoices and transfers stock updates into inventory software.
An end-to-end system for taking, recognising and grading handwritten STEM exams.

An ablation study evaluating various sentiment analysis approaches on the Rotten Tomatoes review dataset, comparing embedding strategies, RNN and transformer architectures, and ensemble techniques.

A MATLAB lab project comparing document image segmentation techniques and stereo disparity methods for 3D reconstruction.

An exploration of essential image enhancement and restoration techniques in MATLAB, covering intensity transformations, spatial filtering with both linear and non-linear kernels, and noise reduction in the frequency domain.

A multimodal emotion recognition system that integrates speech-to-text conversion and deep learning classifiers to detect emotional states from audio and text data.

A network science-based study of DBLP collaboration data revealing structural properties, evolution, and reduction strategies in a large-scale scientist network.

A scalable machine-learning pipeline using dual LightGBM models and outlier detection to predict customer loyalty scores in the Kaggle ELO Merchant Category Recommendation challenge.

A dual-stage deep learning assignment combining speech polarity classification and HDB flat resale price regression.

An implementation of value and policy iteration algorithms that computes optimal policies and utilities in a grid-based maze environment.

An adaptive hybrid-strategy agent for three-player repeated Prisoners’ Dilemma tournaments that balances cooperation and defection through dynamic history analysis.

A reinforcement learning agent implemented with TensorFlow that learns to balance a pole on a moving cart in the OpenAI Gym CartPole-v1 environment.

A user-friendly tool for comparing SQL queries by visualizing execution plan differences and generating plain-English explanations to aid optimization.

A data-driven network traffic analysis tool that processes raw packet logs to reveal communication patterns, protocol distributions, geolocations, and potential security threats.

A market-basket analysis project identifying product affinities and recommending complementary items to boost sales, with additional insights into customer segments and inventory.

A Django-based web trivia game with user authentication, dynamic lifelines, and question management, built as a final exam project for SoftUni.
Discover my tech stack
My technology ecosystem spans multiple domains - from web development to artificial intelligence. Below is a visualization of the technologies I have worked with, organized by category.
6 technologies
6 technologies
10 technologies
5 technologies
11 technologies
5 technologies
6 technologies
Get in touch
Select the appropriate category for your message. I'll receive your message at the corresponding email address.
Professional
For business inquiries and professional opportunities
Personal
For personal matters and general conversations
Educational
For academic and educational-related communications
Usually within 24 hours