Bio
Lovina Dmello is a senior infrastructure software engineer on the Deep Learning Libraries team at NVIDIA, where she works on building and maintaining the infrastructure that powers the NVIDIA deep learning ecosystem. Before joining NVIDIA, Lovina spent four years at Apple on the Apple Payments and Wallets backend team, and three years at Oracle on the Oracle Cloud Infrastructure team. She earned her master's degree in Computer Science from the University of Georgia, where her thesis focused on ransomware classification using machine learning algorithms. Lovina shares her insights through research papers and writing on AI/ML security, agentic AI systems, TensorRT, deep-learning libraries, and infrastructure best practices.
Session (1)
Additional Speaker Details
Using publicly available information we constructed an analysis to help you get a feel for this speaker before deciding to attend their session.
Lovina Dmello brings practitioner-level depth on NVIDIA's inference stack — TensorRT, Dynamo-Triton, Nsight profiling — grounded in the real friction of shipping models to production; her CACM and NVIDIA writing argues that execution bottlenecks, not idea shortages, are what limit AI progress. Attendees focused on moving models from training to reliable, efficient inference will get concrete, tool-specific guidance.
Speaking style
Recent writing (2)
- How to Eliminate Pipeline Friction in AI Model Serving · blog · May 2026
- Why Systems Engineering Increasingly Sets the Pace of AI Progress · blog · May 2026