TL;DR. Build real-time AI to optimize electricity systems for billions, alongside Stanford and ex-Google engineers. Backed by top investors, live across 4 continents, and scaling fast—this is impact at planetary scale.The US base salary range for this full-time position is $150,000 - $300,000 + bonus + equity + benefits. Within the range, individual pay is determined by additional factors, including job-related skills, experience, and relevant education or training. We can share more about the specific salary range during the hiring process. Background. Electricity is undergoing a once-in-a-century reinvention.The grid — the invisible infrastructure powering the world — was designed for a fossil era. It’s now cracking under the weight of extreme weather, data center demand growth, rooftop solar, and decentralized energy. We believe the grid must be made visible, intelligent, and autonomous. At Pravāh, we’re a group of Stanford engineers, physicists, and economists building real-time AI for the electric grid — to make energy more affordable, clean, and resilient for billions. In just weeks since launch, we’ve seen overwhelming traction from utilities and grid operators worldwide. We’re running 7 active pilots across 4 continents, actively deploying our AI systems to improve grid reliability and unlock economic value. From Mumbai to Medellín, we’re building what could become the fastest-growing grid startup in history. Why work with us? AI can transform how we manage energy. Your models will power dispatch decisions in real-time across entire countries, reducing blackouts, emissions, and inequality. This is as direct as impact gets.&nbsp; We’re a small, lean team, and ridiculously ambitious. You’ll be in the trenches with Stanford PhDs, ex-Googlers, and top ML researchers from NASA and IBM. You’ll ship code fast, prototype ML faster, and work directly with utility teams from Delhi to Munich. This is a rare opportunity to join on the ground floor of a funded, mission-driven company with global reach, real customers, and bleeding-edge tech. You’ll shape the culture, roadmap, and architecture of a company tackling one of the biggest challenges of this century. We’re backed by Pear VC and Conviction, two of the top early-stage investors in Silicon Valley. We were selected as one of only 10 startups for Embed, Conviction’s elite incubation program that helped launch breakout companies like Mistral, Reflection, and Pika. With strong early traction and global pilots underway, we’re planning to raise our Seed round this October — which means you’re joining at the inflection point, with an outsized opportunity to shape the company’s trajectory and grow alongside it. What you’ll work on<ul><li>Design and train graph- and transformer-based models for spatiotemporal forecasting of demand, supply, and congestion</li><li>Build physics-informed, probabilistic models to simulate and stabilize volatile power flows</li><li>Develop and deploy real-time ML pipelines that interface with grid telemetry (SCADA, PMUs, AMI)</li><li>Prototype congestion mitigation algorithms using reinforcement learning</li><li>Collaborate directly with power system operators to put your models into production, influencing millions of customers</li><li>Shape our ML infrastructure from scratch — modeling stack, devops, data pipelines, deployment on cloud/edge</li></ul>Who we’re looking forYou're a machine learning engineer who wants to work on urgent, global-scale challenges — not tweak ads. You love going deep on models and owning infrastructure. You can navigate uncertainty, build fast, and are obsessed with impact.What you should have:<ul><li>Strong ML fundamentals: time-series forecasting, sequence models, probabilistic modeling</li><li>5+ years of software/ML engineering experience (or equivalent research + hands-on projects)</li><li>Experience in PyTorch, TensorFlow, or JAX; plus Python, NumPy, Pandas</li><li>Experience with data infrastructure — e.g., Spark, Beam, GCP/AWS pipelines</li><li>Demonstrated ability to ship production-grade systems (bonus if in a high-stakes environment)</li></ul>It’ll be awesome if you had these:<ul><li>Experience with graph neural networks, MARL, or hybrid physics-ML modeling</li><li>Background in energy systems, power engineering, control theory, or signal processing</li><li>Startup experience or prior early-stage roles</li><li>Publications, open source contributions, or prior work on infrastructure/utility-grade ML</li></ul>

Pravāh is a Stanford-founded, AI-powered grid intelligence company on a mission to make electricity cleaner, more affordable, and more reliable for billions. We build real-time forecasting and optimization systems using graph neural networks, transformer models, and reinforcement learning to help utilities manage demand, generation, and congestion. Our customers include grid operators, utilities, and energy traders across emerging and developed markets — from national transmission operators in India and Colombia, to distribution utilities in California, to power trading desks managing renewables at scale. In just weeks since launch, we’ve secured pilots across 4 continents, actively deploying our platform to reduce blackouts, cut emissions, and unlock economic value. Backed by top-tier investors like Pear VC and Conviction, Pravāh is redefining how the world’s most critical infrastructure operates in the age of climate volatility and distributed energy.

Founding ML Engineer @ Pravāh

About the Job

About the Company

Pravāh