Open to opportunities

Varun Bukka

@varunbukka

Message

Dedicated researcher specializing in Machine Learning and Mechatronics.

Germany

Message

What I'm looking for

I am looking for opportunities that foster innovation, collaboration, and growth in AI and machine learning.

I am a dedicated and driven researcher with a strong background in Machine Learning and Mechatronics. My experience includes significant research projects that have culminated in novel techniques for AI model explainability and the application of machine learning to complex data analysis. I am skilled in developing pipelines, utilizing advanced algorithms, and validating models to ensure robustness and applicability.

During my Master Thesis at Forschungszentrum Juelich, I developed an end-to-end ML pipeline for 3D microstructure analysis, achieving 95% accuracy. My research projects at Universitaet Siegen focused on enhancing explainability in deep learning models and improving stock prediction accuracy through financial sentiment analysis. I have also designed GAN models for synthetic data augmentation in healthcare and fine-tuned GPT-3 for personalized education, demonstrating my versatility in applying machine learning across various domains.

Experience

Work history, roles, and key accomplishments

Master Thesis

Forschungszentrum Juelich

Jan 2023 - Sep 2023 (8 months)

Developed an end-to-end ML pipeline for 3D microstructure analysis, integrating data preprocessing, feature extraction, and model deployment. Employed CNNs, hyperparameter tuning (Grid Search CV), and k-fold cross-validation to enhance model robustness. Achieved 95% accuracy, validating predictions against real-world datasets.

GridSearchCV CNN Hyperparameter Tuning K Fold Cross Validation data preprocessing Feature Extraction

Research Assistant

Universitaet Siegen

Mar 2020 - Sep 2023 (3 years 6 months)

Developed NLP models (BERT/RoBERTa) for financial sentiment analysis, improving stock prediction accuracy by 15%. Designed a GAN model for synthetic data augmentation in healthcare, enhancing ML model training and preserving data privacy. Fine-tuned GPT-3 for an adaptive learning platform, increasing student engagement by 20%.

NLP BERT RoBERTa GPT 3 LLM Neo4j Fraud Detection LIME SHAP Power BI Flask

Studienarbeit (Research Project)

Universitaet Siegen

Oct 2021 - Mar 2022 (5 months)

Addressed the trade-off between explainability and performance in deep learning models. Researched ensemble deep learning techniques to balance explainability and performance using XAI methodologies. Applied LIME and SHAP for interpretability on deep learning models, improving trust and adoption.

Deep Learning LIME SHAP Model Interpretability Performance Optimization

Education

Degrees, certifications, and relevant coursework

Universitaet Siegen

Master of Science, Mechatronics

2019 - 2023

Activities and societies: Served as a Research Assistant, working on projects including Financial Sentiment Analysis, Generative AI, LLM for Personalized Education, and Fraud Detection with Explainable AI.

Completed a Master of Science in Mechatronics, focusing on Machine Learning applications. Developed an end-to-end ML pipeline for 3D microstructure analysis and researched explainability in deep learning models using XAI methodologies.