At Sakata INX, I built an in-house LLM-powered desktop AI assistant that automates document creation, web research, data analysis, and file management. It reduced routine task time by 70%.
I also designed the assistant’s semantic tool orchestration and multi-model task routing systems, reducing input-token costs by 91% and average request cost by 60%.
For Sakata INX’s document automation platform, I architected PDF-to-Excel workflows and automated spreadsheet and presentation translation. The work reduced manual effort by 80% and turnaround time by 90%.
Earlier at Sakata INX, I built a multilingual RAG knowledge assistant for HR, IT helpdesk, and department-specific queries. I also developed a TensorFlow PINN project for Kirchhoff-Love plate equations, achieving 1.39% relative error against analytical solutions.

