Sid Valecha

AI/ML Engineer Intern

73 Strings Inc.

June 2025 - August 2025 • New York, NY

• Enhanced LLM-based parsers to streamline data extraction and consolidation from financial documents, reducing processing time by 30% and improving accuracy for AI-driven private equity portfolio and asset management
• Developed and fine-tuned a YOLO-based chart detection model, boosting precision by 12% (to 93%) and recall by 11% (to 94.4%), significantly improving chart classification for financial analytics
• Trained the model on 1,700+ annotated images, achieving a mean average precision of 92.2%, while implementing hyper parameter tuning to reduce false positives and enhance classification reliability
• Explored and proposed innovative applications of Large Language Models for portfolio management, advising the integration of these technologies into the company's product suite, presented in a report.

Full Stack Developer Intern

United Nations (Office of Information and Communications Technology)

May 2025 - June 2025 • Remote

• Developed and implemented a robust file-handling solution for a research repository website, using Node.js, Express, MongoDB, and Multer, streamlining document uploads, storage, and retrieval. Tested and validated API endpoints to ensure efficient backend performance
• Built an LLM-based Retrieval-Augmented Generation (RAG) system for a research portal, utilizing LangChain, FAISS vector storage, and Ollama embeddings to enable fast and context-aware document retrieval. Implemented a document processing pipeline to handle large-scale unstructured data for responses
• Developed a rule-based chatbot for a UNICEF learning management system to resolve user login issues, automating user login issue resolution and reducing support overhead