About the role

The work

• Model electrical, thermal, and energy systems using sparse, noisy data.
• Build agents that plan and execute tasks within equipment and safety limits.
• Create RL loops for multi-objective optimization and long-horizon operations tasks.
• Connect models to facility systems. Test their decisions against real operating conditions. You’ll be helping lead directly on our first data-centers deployments.

Your first 90 days

Build and deploy a power-and-cooling model for a live data-center pod. Connect electrical load, rack temperatures, and cooling equipment behavior to identify bottlenecks, estimate usable capacity, and simulate how operating changes affect power use, temperatures, and equipment limits.

Put that model behind Edviro’s agents so an operator can ask, “Can we add more load here, what needs to change, and what could go wrong?”

You

You’re a strong programmer with depth in machine learning, scientific computing, or systems engineering. You can take an unfamiliar problem, understand the fundamentals, and build a solution without a detailed spec.

You’re interested in RL loops or scaling AI for energy as a proxy for understanding the physical world.

Experience with physical systems, time-series modeling, or controls is useful. We care more about the quality of your work than the length of your résumé.

Send us something technically difficult you’ve built or researched. Tell us what you contributed and what you learned.

Founders

Hursh Shah coauthored & presented at NeurIPS 2024 with a paper on quantum vision transformers for MRI analysis. At 16, he cofounded an AR aircraft-inspection startup, leading its AI and edge-computing systems. Tanuj Siripurapu worked on low-latency neural-network inference on RISC-V FPGA hardware for RF signal classification project at RTX and leads Edviro’s integrations, and deployment engineering. You’ll build directly alongside both founders.

About Edviro

Edviro is building AI that understands, and operates energy infrastructure.

An operator should be able to say, “Reduce this facility’s energy costs without compromising reliability,” and have software understand the equipment, controls, constraints, and tradeoffs well enough to execute a long-horizon decision on critical infrastructure.

To do this we're building the best physics models that understand facilities energy context, combined with other tools that allow our agents to make decisions in the physical world.

Join us to work directly with the founders and own core technology that affects the physical world.