Abhinav P

Data Scientist Ai-ml Engineer @TTEC Digital

Dallas, TX, US

Developed and deployed a Document Question & Answer (Doc-Q AND A) system leveraging Generative AI(Gen AI) for efficient retrieval and generation of answers from large document repositories.Led end-to-end pipeline development, including document ingestion, chunking, embedding generation, vector storage, similarity search, and LLM-based answer generation.Managed ingestion and processing of various document formats (PDFs, DOCs, XLSs, TXTs) using robust chunking mechanisms to improve data handling and retrieval accuracy.Utilized all-MiniLM-L6-v2 model (SBERT- Sentence transformers) to generate high-quality vector embeddings, capturing semantic meaning of document chunks.Employed FAISS (Facebook AI Similarity Search) for efficient indexing and similarity search of document embeddings, enabling rapid retrieval of relevant content.Integrated Llama 2 LLM to generate accurate and contextually relevant answers based on similarity search results.Deployed the system in an AWS environment using AWS S3, Lambda, and EC2 for scalable and efficient deployment using LLMOps pipeline methodologies.Applied model optimization techniques such as pruning and quantization to enhance performance and reduce computational load.Leveraged tools and frameworks including FAISS, transformers, and Langchain to support the system’s functionality.Worked extensively with Python to implement various components of the pipeline.Led a data science project team, managing project timelines, deliverables, and resource allocation to ensure successful project completionDiagnosed and resolved production issues related to model deployment, ensuring minimal downtime and maintaining data integrity.Refined machine learning models in the Doc-Q&A system by analyzing and addressing failed cases in production, improving accuracy and reliabilityApplied critical thinking skills to identify potential data inconsistencies and gaps, ensuring the accuracy and reliability of analytical outputs.

Lindsey Wilson College

Information Technology and Management