Working towards developing a joint frame-pose embedding model integrating 6D camera pose data and video frames to ensure temporally consistent, spatially accurate driving scene generation for autonomous systems. Working under Dr. Sarah Ostadabbas.
• Founded One AI Click to develop a tool aimed at simplifying the fine-tuning process for large language models (LLMs).
• Conducted extensive research and development to create a user-friendly interface and efficient backend for the tool.
• Collaborated with a team of engineers (Srikar Ravi) to implement cutting-edge techniques in LLM fine-tuning.
• Successfully launched a beta version.
• Optimized language model deployment, reducing size by 20% with quantization techniques while maintaining 90% performance.
• Enhanced model specificity using fine-tuning and synthetic data generation.
• Developed a knowledge management system with RAG and corrective agents.
• Created a usability rating system using user interaction data for better UX insights.
• Automated meeting minutes and implemented a chatbot for post-meeting insights.
• Built a Python orchestration solution to optimize workflows.
• Proposed an automated server deployment solution with Python and AWS SDK, improving efficiency.
• Enhanced bancassurance project by designing UI/UX in Figma and developing backend with Spring Boot.
• Implemented frontend in Angular to build data visualizations, improving decision-making efficiency and customer satisfaction.
• Made an animation video to explain oracle bancassurance.
• Additionally, streamlined document templates and improved documentation organization.