Founding software engineer

What we are building

At Elergi we are on a mission to stop the grid being the limit on what the world can build. Almost everything ahead of us runs on electricity: data centres, reindustrialisation, the energy transition. All of it has to reach the grid first, and that is where it stops. In some markets the connection queue holds twice everything already installed. In the Netherlands the queued batteries alone are three times peak load; in Europe's data centre hubs, a project waits seven to ten years for an answer.

Grid connections are about more than just copper. What projects are waiting for is an approval, and someone has to work it out: model the network, run the power flow and stability cases, find where the system fails under load, and size the fix. Manually, one project at a time, while the queue underneath keeps changing and invalidating it. We build agentic AI that does this work on the real physics, across thousands of scenarios, and returns a result an engineer can defend.

We are gathering a team of unusually good people. Between us we have founded companies and taken one from zero to the Nasdaq; mapped and modelled real grids; and worked in consulting, private equity, and AI startups from Stockholm to San Francisco. We like building at a furious pace, and solving problems that are hard.

We are obsessed with our customers, with what would make their work easier and their decisions better. We span several nationalities, we ski, we sing, and we want both the company and its culture to soar.

The role

You would join the founding team and own the machine underneath the product: it takes a grid model and a question, and returns thousands of solved, verifiable cases. Almost none of it exists yet, so the architecture is still open.

What you'd do

• Build the system that turns a grid model and a question into thousands of solved cases, and returns the same answer again and again
• Build the audit trail: every agent decision and tool run recorded, versions pinned, any case replayable years later in front of someone checking it. Here that is a product surface, not plumbing
• Package the whole thing to run inside a utility's own environment: single tenant, their region, nothing pooled
• Own a surface end to end: data model, orchestration, API, and what the engineer actually sees
• Get the loop down to minutes. An overnight batch changes nothing about how planners work
• Take on what only appears at scale: non-convergence, solver crashes, and input data that does not match reality

Requirements

• You have built production systems for several years, including something substantial you designed and shipped yourself
• You are fluent in Python, and in a second language you reach for when Python is too slow
• You have debugged distributed systems under load, and can walk through a failure you caused and fixed
• You want the early version of this job: no spec handed to you, no platform team, users on the phone

Process

1. 01 Intro, 30 minutes What you have built, what we are building, and whether this is worth more of both our time.
2. 02 Deep dive, 1 hour Your own work in detail. What you actually did, what you would do differently.
3. 03 Technical, 1 hour A real problem of ours, worked through together rather than a puzzle from a book.
4. 04 Final, 1 hour The people you would work with day to day, and the questions you still have for us.

We read every application ourselves. If this is the problem you want to work on, we would be glad to meet you.