Shared agentsSecurity and privacy
Keep sensitive information within your organization’s approved data boundaries.
//LeapforIntelligence//Careers
Build the intelligenceenterprises own.
We build domain models, agentic products and the infrastructure under them as one stack.Join the team rebuilding enterprise AI from the ground up.
//Our values
Do whatever it takes to scale intelligence and serve humanity
We work across the whole stack: the data a model learns from, the models that reason over it, the products people use and the hardware it runs on. Most roles here touch more than one layer.
Small teams own problems end to end, from the first experiment to production.
Everything we ship answers to one test: does it make our customers' intelligence more their own?
Momentum
“A working system this week beats a perfect plan next quarter. Ship, measure, and learn from what comes back.”
Ownership
“Own the outcome, not the task. If you see something broken, it's yours until it's fixed.”
Rigor
“Our customers run regulated work. Every number we report has to hold up, so we measure before we claim.”
Craft
“Whether it's a training kernel or a line of copy, build it to last and leave it better than you found it.”
//Open roles
Help build every layer of the stack
Research, engineering, design and field roles across Products, Models and Infra. Every team ships to production, on real enterprise data.
- Models & AlgorithmsFull-timeResearch Scientist, Reasoning ModelsTrain and evaluate domain models that reason over documents, code and clinical records. Own experiments from hypothesis to shipped checkpoint.View role
- Infra & HardwareFull-timeML Infrastructure EngineerBuild the distributed training and inference systems that run our models in isolated, customer-owned environments.View role
- Products & SolutionsFull-timeAgent EngineerDesign the agentic workflows behind Coworker, Developer and Flow, and ship them into the systems our customers already run.View role
- Models & AlgorithmsFull-timeApplied ML Engineer, Fine-tuningFine-tune and align models on proprietary data, with reinforcement learning and evaluation loops that hold up in production.View role
- Infra & HardwareFull-timeInference Performance EngineerProfile and tune kernels, batching and serving paths so domain models answer in milliseconds on dedicated hardware.View role
- Products & SolutionsFull-timeProduct DesignerShape how people work with domain intelligence, from the first prompt to an audited decision, across every DataObserve product.View role
- SolutionsFull-timeForward Deployed EngineerWork inside customer teams in regulated industries to scope, deploy and prove DataObserve systems on their own data.View role
//How we work
Small teams. Whole-stack problems.
Researchers, engineers and designers sit on the same teams and ship to the same customers. A model is only finished when it runs in production, on the customer's infrastructure, under their governance.
We hire people who want to own a problem end to end, and who would rather show a result than describe one.
If that sounds like you, we'd like to hear from you, even if no open role fits exactly.
//Honest limits
This isn't the right fit if…
You want to stay in one layer.
Our problems cross data, models, products and hardware. Most weeks, your work will too.
You need a finished playbook.
Much of what we build hasn't been built before. We write the playbook as we go.
You'd rather not meet the customer.
Every team works with the people who run our systems. Their problems set our priorities.
