CURRENT
Postdoctoral Researcher at TU Berlin
Supervisor: Prof. Fatma Deniz
I work on understanding bilingual language processing in human brain and multilinugal language models.
Visiting Scholar at MaxPlanck Institute for Software Systems (MPI-SWS)
Supervisor: Dr. Mariya Toneva
I worked on understanding the reasons behind improved alignment between language models and the human brain using linguistic probing tasks. My research also explored what types of information contribute to better brain alignment in both text-based and speech-based language models.
EXPERIENCE
Lead Data Scientist at Woundtech Innovative Healthcare Solutions
I worked on AI-driven solutions for automated wound assessment, patient risk of hospitalization, wound image segmentation and NLP-based document processing to enhance clinician decision-making.
- Automated OCR processing using Generative AI & NLP, optimizing invoice and referral processing to run in under 10 seconds, reducing manual intervention across 30+ unique templates for 20+ insurance providers.
- Led the development of hospitalization risk prediction models for critical wounds, integrating wound images and patient attributes (e.g., comorbidities).
- Developed wound segmentation and wound-type classification models using deep learning on wound tissue images (Published at WACV 2020 & WACV 2022), leveraging a dataset of over 2 million (20,00,000) wound images.
- Created a Risk of Readmission model (Presented at CODS-COMAD-2021 Conference) to identify high-risk patients.
- Built and deployed Visits Forecasting models to optimize clinical resource allocation and scheduling.
Data Scientist at Teradata
I worked on NLP, text analytics, machine learning, and graph analytics to enhance Teradata Aster's AI/ML capabilities. My role involved developing word-vector representations, benchmarking classifiers, and optimizing analytical workflows for enterprise-scale data processing.
- Developed NLP models using Word2Vec, GloVe, and fastText for text analytics.
- Enhanced KNIME workflows for Teradata Aster, improving machine learning automation.
- Benchmarked ML models (SVM, Decision Trees, Neural Networks, Linear Regression) on AsterR with standard datasets.
- Implemented text analytics solutions, including NER, POS tagging, and sentiment analysis.
- Worked on dimensionality reduction techniques like PCA for high-dimensional data processing.
Project Engineer at CDAC R&D
I worked on IASF – Intelligent Advisory System for Farmers, an AI-powered Case-Based Reasoning (CBR) system designed to provide automated decision support for farmers. The system stored past queries as cases and leveraged rule-based engines to retrieve solutions for similar queries, significantly reducing response time and improving efficiency.
- Developed an Intelligent Advisory System for farmers using Java and Case-Based Reasoning (CBR).
- Implemented a rule-based engine to automate query resolution and decision support.
