Hello NVIDIA Support Team,
I am an academic researcher/student working on Kairos, a project focused on developing a high-performance, deterministic Discrete Event Simulation (DES) and Agent-Based Modeling (ABM) engine built in Rust.
Project Context:
The engine utilizes a novel Entity Component System (ECS) to manage state and is designed for complex multi-agent simulations. I am currently integrating NVIDIA NIM models into our “Conductor” track system to allow for autonomous, LLM-driven agent reasoning within our deterministic simulation environment. The project is being a polyglot library with a Rust core, and is being adapted for HPC, including GPGPU.
Reason for Request:
My current account is limited to 40 RPM. Because our simulation involves multi-step agent reasoning and parallel agent workflows, a single simulation tick often triggers bursty, high-frequency API calls that immediately exceed this limit. This results in frequent “429 Too Many Requests” errors, making it impossible to validate the deterministic behavior of the AI agents at scale.
Request Details:
-
Account Email: dylan.mordaunt [at] vuw.ac.nz
-
Current Limit: 40 RPM
-
Requested Limit: 200 RPM (or the highest available research tier)
-
Intended Use: Non-commercial academic research, specifically for evaluating agentic reasoning within deterministic ECS-based simulations.
An increase to 200 RPM would allow me to conduct meaningful data collection for our upcoming paper and progress through the planned tracks in our development roadmap.
The library itself is intended to be used on HPC infrastructure. For instance it would be possible to undertake parallel DES / ABM multidimensional modelling for a large population. So any additional suggestions you have on how I could work with NVIDIA or partners on testing this on frontier compute, would be greatly appreciated.
Thank you for your time and for supporting the research community.
Best regards,
Dylan A Mordaunt
Victoria University of Wellington