Software Engineer, Founding Team

In Person, San Francisco

What are we building?

Unlimited is a modern engineering and construction company that builds large-scale physical infrastructure projects like factories, power plants, and industrial facilities. Most companies in this space are >100 years old. These incumbents rely on traditional methods and legacy systems, with incentives often misaligned from their customers, resulting in projects that are slow-moving, risk-averse, tech-resistant, and more expensive and complex than necessary. At the core of what we’re building is Forge - a software platform to unify all engineering and project data into one integrated system with AI as a first class citizen. This is the key to ending the current era of stagnation and reclaiming our ability to shape the physical world.

Who are we?

We are an early stage team of builders. We’ve shipped software products to millions and built infrastructure worth billions. Our current team is Alex, Jordan, Tara, James, Russ, and a couple more chemical and mechanical engineers coming soon. We’re recently raised a large Seed round from top-tier investors (will be announced in a few weeks) to build the best team in the world to tackle this ginormous $9 trillion problem. We believe this is one of the most important and largest opportunities over the next couple decades.

Learn more about us

Your North Star

Design and ship software to accelerate how physical infrastructure projects get built.

If you started today, what would you do?

Build agentic workflows

Work closely with the entire team and our customers to translate real world engineering and project management workflows into AI workflows. You will need strong judgement here and will be responsible for helping shape the requirements of what to build. There will be lots of ambiguity - we’re definitely exploring the unknown and figuring out best practices on the fly. This means we often run into dead ends and need to change our approach, but it’s super satisfying when it works.
Decompose each workflow into automatable components. Set up agents, prompt chains, tools, evaluations, and APIs.
Decide what to automate today vs wait until models improve.
Build generalized services (domain‑specific calculators, file parsing, data validation, etc) to be used by multiple automated workflows.
Build a scalable versioned data model that future products can rely on.

Ship the production grade web app

Collaborate with Jordan to turn designs into production systems. Sometimes you'll get rough design mocks or bullet points, sometimes it will be fully functional web apps that you'll have to rip components out of, etc. You'll contribute your technical perspective and taste to shape the platform.
Turn Russ's scripts or James's Excel models into robust, maintainable product. These will likely be technical in a domain you are not strong in (yet). Eg: thermodynamic calculations to determine heat and mass balances across a system. You’ll need to properly document this for future SWEs who aren’t domain experts and build in resilience and observability as needed for humans and agents.
Build everything we need into the product (dashboard, task management, notifications, 2D flow charts, 3D CAD functionality, etc). Lots of scope and range - you’ll initially be the sole owner of this repo.

Set up evals & observability

Co‑design explicit grading mechanisms (task correctness, completeness, safety, etc) with Russ, James, and a few other domain specific engineers that will join soon. Jordan and Alex will also be involved in this. Building a killer set of evals is arguably the most important thing we do in the next 6-12 months.
Enable Russ and other domain engineers on our team to rapidly iterate through new tests and optimizations to automate workflows.

Select and implement tools and standards

Full stack for production web app.
Jordan will likely provide most of the UI in react components via V0/cursor. Ideally you help Jordan get even more leverage to cut down the need for additional front end work and establish super low friction handoff. Jordan is a software engineer by education, but an incredibly talented product designer (this is where he should spend his time).
Try new frameworks, DBs, APIs, and other tools, to build the product and communicate whats possible back to the rest of the cross-discipline eng team. We want you to build strong opinions that are likely going to shift as we all learn more and tech advances.
Set standards for engineering at Unlimited (code quality, reviews, interviewing process, etc)

At the end of the day, you will be the first dedicated software engineer and responsible for wearing a ton of hats. This is not a straight forward SaaS web app. It will require you working in a fully deployed motion with customers and closly with our cross discipline team to contribute to set the requirements. You’ll set the tone for the team as we scale and have incredible influence on the direction of this company.

‍

Core Competencies

Must‑haves

Demonstrated ownership of backend / platform services end‑to‑end, architected, shipped, and supported in production.
Bias to act, learn, iterate. We find answers through rapid cycles, not endless theorizing.
Excellent communicator across software & non‑software engineering roles.
Great technical design judgment.
Strong data‑modeling instincts (schemas, migrations, indexing).
Hands‑on experience with AI/LLM tooling (e.g., prompt chains, evaluation harnesses, agent frameworks) and a demonstrated curiosity to keep experimenting with new capabilities.

Nice‑to‑haves

Experience with CAD and simulation tools
Lots of hands‑on with new AI tools (you’re a tinkerer)
Infra / construction‑tech / digital‑twin / physical engineering domain exposure

Not a good fit

Background in primarily research vs. shipping to prod.
Are not excited and hungry to explore AI (i.e. if you haven’t been following along and playing with new tools)

How you’ll know you're succeeding in 60 days

Increase the prototype → prod cycle time and allow Russ + domain engineers to scope out more workflows.
We have a working product that James and our domain engineers can use to accelerate deliverables for our customers. We’re product to demo this to customers.
Russ + domain engineers say you make their lives way easier because of your contribution. For example:
- the product is reliably working and they are able to us it the same way a future customer would
- on your suggestion, you built internal tools for them because you knew they would be helpful
- you offer good insights on what’s possible from an AI perspective and have opinions on how to design workflows, etc
A practical evaluation harness is in place. Automated tests run on demand (push to prod, testing in dev, optimizing prompt/model selection)
You’re becoming more familiar with mechanical, chemical, electrical engineering concepts. You’re proactive in your learning, or at least asking questions to help see around corners.
You create your own list of high value improvements to make to our workflows, verification, and evals.
You are quick to respond on slack and have a really good working relationships with other non-software engineers on the team. They proactively reach out to you and trust the answers you give them.
You operate with minimal oversight. surface risks early, set deadlines, meet them.

How you’ll know you're failing in 60 days

Need to hire another software engineer to implement something you can’t do (too technical, to basic, etc)
You aren’t ramping up on basic non-software engineering concepts and require frequent reexplaining of foundational concepts, limiting the speed of useful collaboration.
You are dogmatic in how we need to do something but haven’t been obviously proven right.
There hasn’t been any measured value of the automation yet and don’t have a ton of iterations to at least show further our understanding of what doesn’t work.
You aren’t developing more intuition around complexity and what’s possible to automate vs what we should wait to do until later.
You’re not super easy to work with (this is very important on a cross-discipline team where no one knows everything, and collaboration requires the willingness to look dumb and be wrong).
Need step‑by‑step direction; wait for tickets instead of hunting for impact.
Systems Engineer remains blocked or has to build temporary tooling themselves.
Unused evaluation or verification framework; leads to silent failures.
There are few running services in prod. Most work remains in local branches or notebooks.
No forward looking software roadmap. You are entirely reactive to tickets that have been assigned to you.
You made many poor architectural decisions that now require the majority of your time spent maintaining systems or dealing with preventable bugs.