AI Practices Intern
As an AI Practices Intern at Affine, I contributed to building a Multimodal RAG pipeline that ingested, normalized, and chunked large volumes of documents and tabular data. My work included generating dense vector embeddings and orchestrating LLM-driven retreivers and ranking mechanisms in conversational workloads. I implemented document parsers and audio transcription pipelines to process, annotate, and prepare rich labeled data for search and retrieval tasks. • Built automated pipelines normalizing PDFs, CSVs, OCR images, and transcripts into vectorized data chunks for AI retrieval. • Generated embeddings using Sentence Transformers and deployed them for text understanding and search tasks. • Implemented and labeled conversational data and ranking for LLM systems, supporting scalable AI use cases. • Managed processing, labeling, and evaluation of over 10,000 minutes of audio and 85,000 documents monthly.