Working on Large Language Model (LLMs) use cases in Legal, Finance domain.Leveraging open source Large Language models like FlanT5XXL, Falcon, Llama2, Mistral and their commercial counterparts OpenAI gpt-3.5, gpt-4, Google Text Bison to solve business problems.Worked to LLM finetuning (PEFT - LoRA) and other Transformer based models in general.Worked on text data preprocessing for LLMs. Leveraging Langchain, Vector Stores and prompt Engineering to effectively use LLMs and solve business problems.Build Retrieval Augmented Generation (RAG) pipelines to extract information from commercial lease documents.Worked on business problem faced by big financial and retail tech company to automate its complex manual workflow using natural language processing.Worked on entity information extraction and line of business identification for top insurance company.worked on invoice and remittance processing using natural language processing and computer vision.Responsibilities:Delivering solutions using Natural language processing, machine learning and computer vision.Analyzing and preprocessing the data to train AI models.Experimenting with state of the art models and using it to solve the business problems. Collaborating with business experts for business understanding and improvements in the ML models.Continuously improving the asset so as to adapt to different client with similar problems.Built and presented proof of concepts(PoC) and Proof of Value(PoV) using LLMs & conventional NLP.

Working on Large Language Model (LLMs) use cases in Legal, Finance domain.Leveraging open source Large Language models like FlanT5XXL, Falcon, Llama2, Mistral and their commercial counterparts OpenAI gpt-3.5, gpt-4, Google Text Bison to solve business problems.Worked to LLM finetuning (PEFT - LoRA) and other Transformer based models in general.Worked on text data preprocessing for LLMs. Leveraging Langchain, Vector Stores and prompt Engineering to effectively use LLMs and solve business problems.Build Retrieval Augmented Generation (RAG) pipelines to extract information from commercial lease documents.Worked on business problem faced by big financial and retail tech company to automate its complex manual workflow using natural language processing.Worked on entity information extraction and line of business identification for top insurance company.worked on invoice and remittance processing using natural language processing and computer vision.Responsibilities:Delivering solutions using Natural language processing, machine learning and computer vision.Analyzing and preprocessing the data to train AI models.Experimenting with state of the art models and using it to solve the business problems. Collaborating with business experts for business understanding and improvements in the ML models.Continuously improving the asset so as to adapt to different client with similar problems.Built and presented proof of concepts(PoC) and Proof of Value(PoV) using LLMs & conventional NLP.

Key contributions include:Text SummarizationDeveloped a Seq2Seq model with LSTM and attention mechanisms to generate concise, meaningful, and fluent summaries, adaptable to various document-summary pairs. The model captures essential context vectors, ensuring key information is preserved.BERT for NLP TasksImplemented BERT-based models for various natural language processing tasks such as sentiment analysis, named entity recognition, and text classification. Fine-tuned pre-trained BERT models to improve accuracy on specific datasets, leveraging transfer learning to reduce training time while enhancing performance.Face RecognitionCreated a face recognition solution using a FaceNet model with a ResNet-v1 backbone for face embedding extraction. The SVM classifier delivered high accuracy, even with limited data, tailored for organizational needs.MixIT Software DevelopmentLed the development of the computational software (MixIT) on both frontend and backend. Created new utilities and features by analyzing business requirements and client feedback. Involved in unit testing, bug fixation, installation package creation, and comprehensive end-to-end documentation.

Responsibilities:Conducted profiling of potential recruiters for the Mechanical Engineering Department at IIT Guwahati, aligning student skill sets with industry requirements.Engaged with recruiters to effectively communicate the value and capabilities of our students, successfully securing their participation in the placement process.Managed end-to-end placement activities, including organizing Pre-Placement Talks, coordinating Written Tests, facilitating Group Discussions, and overseeing Interviews.Acted as the primary liaison between students and recruiters, ensuring smooth communication and addressing any concerns from both parties.