• Technical Design: Expertly designed and implemented High-Level Designs and Low-Level Designs for various modules, ensuring robust and scalable architectures.• Chat History Management: Developed a user chat history management system using Azure Cosmos DB, optimizing data retrieval and enhancing user access. • Voice-to-Text Transcription, Summary Verification: Developed a solution that generates transcripts from voice recordings of conversations, with a LangGraph Agentic flow to summarize key discussion points and verify the summary against specific parameters. An automated LLM powered internet search then finds potential solutions, and the final output is a comprehensive PDF containing the conversation summary, main problem, and researched solutions.• RAG Pipeline Development: Implemented a Retrieval-Augmented Generation (RAG) pipeline using LangChain powered by GPT-4, enabling precise extraction of relevant information from large custom documents with thousands of pages.• Dynamic Report Generation: Built a dynamic reporting pipeline that generates customized reports based on user-bot conversations and user custom documents. The solution had Generative AI support to extract relevant information and generate a PDF report focused on key parameters.• User Management System: Designed and implemented a comprehensive user management module supporting login, registration, and role management, ensuring secure and efficient user access control.• Developed and delivered the tool with robust backend deployed on Azure cloud.

• Machine Translation System with Falcon, Bloom, and GPT-3.5: Developed a robust machine translation system utilizing Falcon, Bloom, and GPT-3.5 to achieve high accuracy and context-awareness across multiple languages. The efforts involved fine-tuning of open source LLMs for language specific translation tasks. This solution enabled seamless communication and information exchange by preserving the nuances of language, even in complex or technical texts.• Custom Q&A System Using RAG with LangChain and Haystack: Built a Retrieval-Augmented Generation (RAG) system leveraging LangChain, Chroma, and Haystack. This solution enables dynamic interactions with a custom Q&A database, allowing users to ask specific questions and receive accurate, context-driven responses, making it ideal for customer support or knowledge management.• Enhanced Data Retrieval with Mistral and GPT-3.5: Integrated models like Mistral and GPT-3.5 into a data retrieval pipeline, optimizing response accuracy and relevance. By using these models, I developed a system that can handle complex queries and retrieve information from large datasets efficiently, aiding in research, data analysis, and decision-making processes.

• Developing and implementing advanced NLP algorithms and models for text classification, Neural Machine Translation, Aspect based Semantic Analysis and NER.• Experienced working with Generative AI Frameworks like Langchain• Experienced working with LLM Models like GPT, Falcon, Bloom, Mistral, etc.• Preprocessing and refining vast textual datasets to unveil valuable insights and patterns.• Collaborating with cross-functional teams to translate business requirements into NLP-driven solutions.• Fine-tuning and optimizing NLP models using state-of-the-art frameworks like TensorFlow, PyTorch, or Transformers.• Conducting in-depth exploratory data analysis to unravel the hidden potential within textual data.• Evaluating, benchmarking, and deploying cutting-edge NLP techniques for superior accuracy and efficiency.• Staying at the forefront of NLP research to integrate groundbreaking advancements into practical applications.• Troubleshooting and fine-tuning NLP systems to enhance performance and user experience.• Documenting research findings, methodologies, and codebase to foster knowledge sharing and maintain robust documentation standards..