Monishka Naidu

Built a RAG system using Gemini LLM to streamline knowledge access, reducing information retrieval time for managers and workers by 40%. Engineered OCR pipelines via Google Document AI for 30+ languages, saving 100+ manual hours monthly in data entry. Researched, compared, and integrated Google APIs for Translation, Speech-to-Text, and Text-to-Speech. Automated testing for 40 target languages, ensuring 95-99% validation accuracy before production deployment. Collaborated with cross-functional teams to assess scalability and real-world adoption of GenAI solutions.

Developed ML solutions for the government sector, improving fraud detection and image processing accuracy by 70%. Designed and implemented image optimization pipelines using OpenCV, improving image quality and downstream AI model performance. Applied computer vision techniques for face recognition image processing in production environments. Presented technical findings to the client, leading to a data-driven decision to descope the signature verification feature due to natural human signature variability and long-term cost considerations, leading to a 15% reduction in project scope for cost-efficiency. Integrated AI models using Flask APIs and deployed services through CI/CD pipelines to ensure scalability and reliability.