Software Infrastructure Engineer

Company Overview

Reflex Robotics is building affordable ($10k) humanoid robots to automate dangerous and repetitive tasks in manufacturing / logistics. We envision a future where everyone has their own “RobotGPT” that can do the boring stuff and help them focus on more fulfilling work.

We are a two-year old startup backed by Khosla Ventures, with $60m/year of revenue lined up pending successful pilots with e-commerce warehouses in 2024. 

How does it work?

Our robots are designed & built in-house by an engineering team that led development of the Stretch robot at Boston Dynamics & key parts of the Model S,X,Y production line at Tesla. Reflex robots are high-performance, low-inertia, and designed for low-cost manufacturing.

We've also built a tele-operation system by which someone far away can basically “play a video game” and control the robots in real-time. This lets us do a LOT of tasks on day one, increase autonomy over time, and still have remote operators as needed to ensure that our robots don't make mistakes.

Key company beliefs

• Having high quality proprietary data is the key to building generational AI companies. Tesla FSD has 1B+ miles of driver data & ChatGPT has the whole internet - but there is no good dataset in robotics.

• It is pointless to focus on overly academic things (e.g. maximal torque density, 0.1mm actuator precision, getting bipedal robots to walk) instead of actual usefulness to customers.

• Human prosperity increases super-linearly with increased automation. Freedom from dull & repetitive tasks allows us to focus on more rewarding pursuits.

• An insane work ethic is necessary for outsized success.

Engineering At Reflex:

Reflex is uncompromising on building a world-class engineering team. Our team has shipped material handling robots at Boston Dynamics, packaging and home robots at Amazon, drones deployed in Ukraine at Skydio, and nm-scale lithography control at ASML.

We strongly believe in providing engineers with the autonomy to make decisions and see them through. This has allowed each person on our engineering team to be independently obsessed with improving the product and has fostered a work culture that is every bit as intense as it is rewarding.

The Role:

There’s a lot of work that goes into deploying a fleet of robots that can reliably do useful work at multiple customer sites shift after shift, day after day. As we deploy hundreds of robots over the next year, we need to double down on building great infrastructure and internal tooling to maximize the impact of our engineering efforts.

This involves building and managing:

• Internal visualization tools (e.g. using Rerun) to debug issues and enhance system observability

• Tools to auto-upload new ML models to our bots each day, using benchmarks from learned simulators (world models) to check for regressions

• Data ingestion pipelines and cloud services to host all of our robot data from the field to enable training our models

• Remote monitoring and diagnostics tools and dashboards to keep us informed of the status of our fleet

• Deployment infrastructure like feature flags and progressive roll-outs to ensure smooth deployment of new software and firmware updates

This role would be a great fit if you:

• have shipped real robots or complex software+hardware products in the past

• pride yourself on writing robust code that the team can rely on for years

• understand the industry best-practices and also how to best apply them under changing needs

We primarily use C++, with some Python for prototyping. We don’t use ROS.

You’d be joining at a time when there’s working hardware, working remote-operation software, and a qualified set of pilot customers – the company is de-risked enough to see the hazy outlines of success, but not so large that there’s no remaining upside.

Come join us!