Leading data science initiatives focused on large language model (LLM) integration and intelligent document understanding solutions.- Optimized LLM inference performance through innovative techniques leveraging modified KV Cache and frameworks like vLLM, TensorRT, and Llamacpp.- Developed high-accuracy machine learning models for address quality assessment and scoring.- Championed cross-team collaboration to explore and implement LLM-based solutions across the organization.- Leading the development of a universal PDF parser for efficient extraction of information from various document types (payslips, resumes, ID cards etc.).

Spearheaded innovation in Natural Language Processing (NLP) and Generative AI (GenAI) applications.- Developed and implemented advanced NLP models for name classification and address segmentation tasks, achieving significant accuracy improvements and efficiency gains.- Pioneered research into GenAI applications within the company, exploring its potential for various use cases and compatibility with OpenLLMs and proprietary LLMs (GPT, Gemini).- Conducted performance analysis of Vector Databases, embedding methods, indexing techniques, and neighboring algorithms to optimize information retrieval for NLP tasks.- Headed efforts to introduce GenAI capabilities across different teams within the organization.- Initiated the development of a Universal PDF Parser to facilitate comprehensive document understanding.

Led a team of data scientists and software engineers in developing advanced information extraction solutions for unstructured documents, specifically focusing on legal documents.- Designed and implemented Natural Language Processing (NLP) models for Named Entity Recognition (NER) and entity relationship extraction.- Developed rule-based and machine learning models for information extraction tasks, including address parsing, attribute classification, and event detection.- Achieved significant improvements in data extraction accuracy, such as increasing address matching accuracy by 10%.- Built and integrated an Address Quality API to enhance data processing pipelines.- Utilized deep learning techniques (CNN, BiLSTM, CRF) to develop an address segmentation model with high accuracy and low latency.

Focused on building automated data extraction and processing solutions using Natural Language Processing (NLP) techniques.- Developed and deployed crawlers to gather data from various sources.- Performed large-scale data analysis to identify patterns and improve data quality.- Built rule-based and machine learning models for tasks like address parsing, name matching, and entity recognition.- Achieved significant accuracy improvements in data processing tasks.- Contributed to various NLP projects through proof-of-concepts, integrations, and analysis.- Managed a team of data scientists and software engineers.