Aravind Ryali

Master Thesis – Model Compression of LLM at Elektrobit

Developed a local large language model 300 times smaller than commercial models with 92% accuracy by implementing knowledge distillation, few-shot prompting, fine-tuning, and prompt engineering techniques. Containerized the application with Docker and set up automated deployment via CI/CD pipelines. Deployed the model as a production-ready AI assistant for vehicle dashboards, optimizing resource usage and real-time performance.

Data Analyst (Working Student) at Siemens Healthineers

Performed ETL processes on datasets exceeding 2 million records to ensure data consistency and scalability. Designed and deployed over 12 interactive dashboards using Power BI and Azure, improving stakeholder reporting efficiency by 60% and enabling real-time project tracking.

Data Analyst (Working Student) at CheckMyBus

Executed ELT operations and developed 5+ interactive dashboards with Power BI to support data-driven decision-making. Managed data quality on datasets exceeding 100,000 records within an in-house content management system. Collaborated with cross-functional teams to ensure data models and reporting structures aligned with key business objectives.

AI Engineer (Master Thesis) at Elektrobit

Built a compact local voice assistant LLM with 92% accuracy using knowledge distillation, few-shot prompting, fine-tuning, and prompt engineering. Containerized Docker and implemented automated deployment through CI/CD pipelines. Deployed the LLM as a production-ready AI voice assistant for vehicle dashboards.

Data Analyst (Working Student) at Siemens Healthineers

Performed ETL on datasets exceeding 2 million records, ensuring data consistency and scalability. Designed and deployed 12+ interactive dashboards using PowerBI and Azure, improving stakeholder reporting efficiency by 60% and enabling real-time project tracking.