Applied Scientist

Rabot builds vision AI for warehouse packing operations. Our systems observe physical processes through cameras, run inference on edge devices, and deliver real-time feedback to human operators. The technical surface spans computer vision, real-time embedded systems, cloud infrastructure, and human-facing software.

We're venture-backed, deployed with paying customers, and partnered with major industry players. The engineering problems are real and the systems run in production, not in a lab.

The problem

Our product sits at the intersection of several hard systems: cameras and optics in uncontrolled environments, AI models running on constrained edge hardware, real-time data pipelines, cloud-scale analytics, and software interfaces for non-technical users. These systems interact in ways that are difficult to reason about without formal tools.

We're looking for someone who can think about these systems at a level of abstraction above the code. Someone who sees architecture problems as problems in combinatorics or graph theory. Someone who models data flow the way a physicist models energy flow. Someone who can identify the fundamental constraints in a system, not just the implementation bottlenecks.

AI tools have changed what's possible here. A person with deep theoretical training and strong AI fluency can now architect a system, validate it formally, and implement it, all without needing a team of specialists. We're hiring for that person.

What You'd Work On

• Analyze and redesign the abstractions across our technical stack. Internal tools, customer-facing software, edge systems, AI models. Find the unifying structures.
• Model system behavior formally where it matters. Latency bounds, throughput limits, failure modes, scaling properties. Use the right mathematical framework for the problem.
• Work across teams as the person who sees the whole system. Translate between the hardware engineer thinking about device constraints and the software engineer thinking about user experience.
• Identify where AI models can replace heuristics or manual processes, both in the product and in how we build it.
• Use AI tools as a core part of your workflow. For implementation, for exploration, for validation. We expect you to be fluent.
• Ship. Theoretical elegance matters, but so does production code. You'll have AI tools to help bridge the gap, but the work has to reach customers.

• You have deep training in abstract reasoning. Mathematics, theoretical physics, theoretical computer science, or a related discipline. PhD preferred, but what matters is the depth of thinking, not the credential.
• You can formalize problems. When you see a messy engineering challenge, your instinct is to find the right abstraction, define the constraints precisely, and reason about the solution space before writing code.
• You're AI-fluent. You use AI tools every day as thinking partners and implementation accelerators. You see them as what they are: tools that let one person with deep understanding do what used to require a team.
• You can communicate with engineers. You don't just prove things; you explain them in ways that change how people build software.
• You ship. You may not be the fastest coder on the team, but between your understanding and AI tools, your work reaches production.
• You're drawn to hard problems in messy domains. Warehouses are not clean rooms. The interesting part is making rigorous systems work in uncontrolled environments.

• Experience with computer vision, perception systems, or signal processing.
• Background in optimization, control theory, queueing theory, or information theory applied to real systems.
• Familiarity with edge computing constraints: limited memory, power, compute.
• Experience deploying AI/ML models in production (not just training them).
• Publications or research output that demonstrates original technical thinking.
• You've worked in industry before and understand the difference between a proof and a product.

• Base salary plus equity. A real stake in the company.
• Hard problems at the intersection of AI, physical systems, and software.
• A small team where your thinking directly shapes the product and architecture.
• Direct access to founders. The CEO holds a PhD in Electrical Engineering from UT Arlington, where his research proved stability of neural network-based real-time controllers using the Lyapunov method, analogous to classical proofs of Kalman filter stability. He speaks your language.
• The problem domain has hard theoretical components drawing from topology, Lie algebra, control theory, and information theory. This is not a company where theoretical depth goes unappreciated.
• AI tools and a culture that uses them seriously.

We want someone who bets on themselves. If you're optimizing purely for guaranteed base, this probably isn't the right fit. If you want to apply deep technical thinking to a real product at a company where you own a meaningful piece of the outcome, this structure works.

How To Apply

Send us two things:

• A piece of technical work you're proud of. A paper, a system you designed, a proof, a project. Something that shows how you think, not just what you built.
• You observe a system where throughput degrades non-linearly as load increases, but no single component is saturated. What frameworks would you reach for to diagnose this? How would you formalize the problem? Keep it under a page.