- Spearheaded the development of an end-to-end pipeline for timeline generation in the legal domain, leveraging Large Language Models (LLMs) , increasing the revenue of the company by 5% and decreased manual intervention by 70 % with this new feature.- Specialized in agentic development, designing AI agents that improved autonomous decision-making and task execution by 20%, significantly enhancing operational efficiency across diverse applications in dynamic environments.-… Show more - Spearheaded the development of an end-to-end pipeline for timeline generation in the legal domain, leveraging Large Language Models (LLMs) , increasing the revenue of the company by 5% and decreased manual intervention by 70 % with this new feature.- Specialized in agentic development, designing AI agents that improved autonomous decision-making and task execution by 20%, significantly enhancing operational efficiency across diverse applications in dynamic environments.- Developed a pipeline for entity normalization via knowledge graphs, addressing challenges like name variations and synonym handling, improving the accuracy of entity extraction by 12%.- Utilised Generative AI, Prompt Engineering, and Deep Learning to build robust solutions for Retrieval-Augmented Generation (RAG) tasks, leading to a 10% improvement in document retrieval accuracy.- Enhanced proficiency in Natural Language Processing (NLP) to drive innovation in LegalTech. Show less

Integrated Pipeline: Designed and developed a comprehensive pipeline that encompasses data acquisition, cleansing, preprocessing, model training, and deployment. This pipeline enables human activity recognition by analyzing micro-Doppler signatures captured by the radar.Deep Learning Architectures: Implemented state-of-the-art deep learning architectures such as Vision Transformers, Convmixer, and Encoder-Decoder-based models. These architectures were used to analyze segmented… Show more Integrated Pipeline: Designed and developed a comprehensive pipeline that encompasses data acquisition, cleansing, preprocessing, model training, and deployment. This pipeline enables human activity recognition by analyzing micro-Doppler signatures captured by the radar.Deep Learning Architectures: Implemented state-of-the-art deep learning architectures such as Vision Transformers, Convmixer, and Encoder-Decoder-based models. These architectures were used to analyze segmented spectrograms with an impressive accuracy of 98%.Cross Frequency Adaptation: Addressed the Cross Frequency Adaptation problem between 77GHz and 24GHz in Radar Systems. By utilizing Few Shot Learning Algorithms (FSL), you successfully improved the performance and accuracy of the system.Cutting-Edge DL-Based Pipeline: My work resulted in the development of a cutting-edge deep learning-based pipeline for human activity recognition using radar. This pipeline revolutionized the traditional approach and achieved optimal performance.Signal Processing: Worked on various aspects of signal processing, including slow time and fast-time signal processing, MTI filters, range and velocity estimation, generation of RD maps, and spectrogram analysis. Show less

• Designed and Developed an automated fraud detection system for the client by leveraging Python and Scikit-Learn to create a robust machine learning model, achieving an impressive accuracy rate of over 95 % in identifying fraudulent cheques.• Continuously evaluated and monitored the performance of ML models on incoming data streams of customers from UAT, and SIT teams, proactively detecting and addressing any data drift to maintain high accuracy and efficiency.•… Show more • Designed and Developed an automated fraud detection system for the client by leveraging Python and Scikit-Learn to create a robust machine learning model, achieving an impressive accuracy rate of over 95 % in identifying fraudulent cheques.• Continuously evaluated and monitored the performance of ML models on incoming data streams of customers from UAT, and SIT teams, proactively detecting and addressing any data drift to maintain high accuracy and efficiency.• Effectively led a Data Analysis and Visualization project using Quality Centre/JIRA Reports, ensuring adherence to strict deadlines and robust quality checks to meet the requirements of stakeholders and ensure project success.• Designed and implemented an advanced machine learning model for customer segmentation based on the set of fields provided by the client, which wasthen used to effectively target specific customer segments for loans and other privileges.• Collaborated closely with the client to deliver actionable insights and implemented machine learning-based automation solutions for cheque-clearing systems, and thus optimized business process.• Ownership of data and automation tasks for the team, and ensured accurate analysis and robust audit control checks.• Executed and debugged ETL Graphs in Abinitio for successfully loading data from source to target.• Developed and executed SQL queries to validate data integrity and ensure that data is being transformed according to business requirements Show less