Hyderabad, Telangana, India
Summary
Designed and implemented a comprehensive backend system for evaluating, fine-tuning, and monitoring Large Language Models (LLMs), integrating real-time metrics and system performance tracking to optimize model development.
Highlights
Engineered and deployed robust RESTful APIs with Python and FastAPI, enabling real-time evaluation of LLM models by integrating key metrics like Cosine Similarity, ROUGE, and BLEU, enhancing evaluation speed by 15%.
Developed and integrated real-time monitoring APIs with Prometheus for GPU, CPU, and RAM utilization, proactively identifying and resolving resource bottlenecks during high-load LLM training sessions, ensuring 99.9% system uptime.
Contributed significantly to the architectural design of AI-powered backend systems, optimizing API performance and data pipelines, which enhanced LLM solution scalability by 25% and improved processing efficiency.
Created and deployed interactive Grafana dashboards and Matplotlib visualizations for real-time monitoring of LLM evaluation and fine-tuning metrics, enhancing stakeholder insights into model accuracy and resource utilization by 40%.
Managed Agile project workflows using JIRA for sprint planning and issue tracking, and automated JIRA-Slack integrations, improving team coordination and communication efficiency by 20%.