Software Engineer

Location: SF Bay Area - Hybrid

Compensation: $140k - $170k + .05-.2% equity

Engineering Job Description
Factory safety practices haven’t changed since the 50s -- and it shows. Despite automation technology and advances in global supply chain management, worker injury costs have increased yearly and the largest fraction of injury still comes from material handling and assembly tasks. At TuMeke we’re leveraging automation to keep factories safe and cut down on injury rates. Our first product uses computer vision to perform safety inspections on factory workstations automatically.

We are looking to hire a full stack engineer to join our twenty person team. The ideal candidate is someone with deep expertise in Python, Javascript (ReactJS), and Kubernetes (with more emphasis toward front end engineering 70/30 split). This person is also comfortable getting their hands dirty understanding customer needs & setting product direction. They will shape our engineering culture and help cement TuMeke’s reputation as a world class ML organization.

If you join us, you’ll be an early team member in helping guide:
The culture of our organization
Who we hire
The focus of our products
Engineering best practices

Challenges you’ll work on:

• Implement frontend UI for critical new features that help reduce injury risk in factories.
• Improve the time it takes to re-encode user videos in a web playable format.
• Develop a cogent local devops stack
• Optimize processing times by starting processing loop as videos upload into S3
• Implement an intelligent caching system to improve the speed of critical database operations

Problems our backend team has worked on (successfully!) so far:

• Integrated pose estimation model inference directly into browser app and iOS app.
• Increased paid user conversion rates by an order of magnitude by implementing more intuitive UI paths.
• Implement a novel video DRM protection scheme at ~1% of the cost of commercial alternatives.
• Implementing AWS infra in-code using Terraform.

Successful candidates:

• Have 3 years of experience with Python and Javascript.
• Have built robust CI/CD pipelines, and are familiar with React and React Native
• Are willing to serve on-call periodically
• Are excited to create a new product in a space that hasn’t seen innovation in a generation.
• Supportive—especially when teammates are faced with new challenges.
• Feel a great sense of accountability to each other.
• Uphold best practices in engineering, security, and design.

FAQ
Who cares about factory safety?
Factory injuries are an incredibly expensive problem -- last year US manufacturers spent $60bln on healthcare costs for their injured workers. The remarkable thing is that this cost isn’t from freak accidents like a bandsaw going haywire (although that does happen); it’s from soft tissue injuries that occur in the day-to-day of just doing your job. That means these injuries can be prevented.

Companies try all sorts of things to reduce their injury rates, from training sessions to even hiring physical therapists to watch workers do their jobs. The majority of these techniques are incredibly manual and time intensive procedures, so a lot of companies don't bother. This means there are millions of people being injured worldwide that don’t need to be.

What about robots taking everyone’s job?
There’s no doubt that automation is eating away at manufacturing work -- but not as much as some politicians/journalists would have you believe. A simple example: the most common job at Amazon (with all of its financial/human capital) is still in the warehouse. Even something as simple as box picking hasn’t been automated yet and humans are doing the bulk of the work -- and getting unnecessarily injured while they are at it. Skilled manual labor will be an integral part of the global economy for at least the next 20 years.

As another data point, Amazon announced recently that it was investing $300 million to reduce the prevalence of musculoskeletal disorders (MSDs) in their distribution centers. They were going to do this by hiring teams of physical therapists and coaches to re-train employees and monitor whether jobs are done in a biomechanically correct way. This company is purportedly at the forefront of automation, and the best solution they’ve found to MSDs is just to hire a bunch of people to coach their workers.

Who will I be working with?
The three founders Diwakar, Riley, and Zach, Alan Hedge, Mark, Dr. Jun He, Balaji, Sofia, and Mark. Dr. Hedge is a titan in the field of industrial design and is a leading expert on worker safety. Dr. He was the head of the ML department at Nanjing University, and is an expert in human pose estimation and activity recognition. Our new backend engineer will get to work closely with him to bring to life the theoretical work he’s done his entire career.

How does this product technically work?
We use a neural net to extract joint information from a video and use some basic signal processing techniques to reduce noise in the data and discard joint data we think will not be useful for tracking industrial jobs. Then we go through standard medical diagnostic tools to estimate the risk of a job from that joint information.

What state is the product in?
We have a product being used by factories right now, and we have early evidence that our solution reduces injuries for customers. This group includes some of the largest/most respected manufacturers out there -- happy to share more info during an interview.

How does TuMeke make money?
We charge our customers (insurance companies and manufacturing companies) a monthly subscription per specialist that uses our tool. We’ll soon cross $3m ARR with our MVP product.

How big is the market for this?
We currently charge a monthly subscription per safety specialist that has a seat on our platform. There are ~840k safety specialists in our target markets so if we multiply through by our monthly price we get a starting market of: 840k * $250 * 12 = $2.5B.

Eventually we’ll charge a monthly subscription per worker that’s assessed on our platform. Our competitors that make wearable devices to track worker posture charge ~$20-25/mo/employee, so we believe we can charge a fraction of that and still enjoy software margins (thanks to the algos you’ll help build! :)). There are 201M industrial workers in our target market, and if we charge $15/worker/month, that’s a $20B annual market! Definitely venture scale.

What’s the interview process like?
Initially we’ll do a quick phone call to make sure that we’re on the same page. After that, our process is:
Shorter 1-hr technical interview with problems that we’ve faced at TuMeke before. No trick/puzzle questions.
(Paid) Take-home project that should take 2 days to complete.
Do a formal interview with the team. Should take 2-3 hours.
Reference checks with 3 other people you’ve worked with professionally.

Have you raised money?
Yes! We’ve raised more than ten million in venture dollars, and we’d be happy to share details on the phone :) Our investors include Intel Capital, GSRV, Tuesday Capital, and OVOFund.