Carbic builds a new type of sensor and software system to monitor fluid infrastructure around the world for leaks, theft, spills, and performance. Fleets of our proprietary lightweight hardware generate raw ultrasonic data and then send that data directly to our server via cellular/satellite; we then use a mix of physics equations, signal processing, and anomaly detection to determine the behavior of fluid (volume and speed) in pipes.
Comparable sensors can cost $200k+ apiece; we're selling on a SaaS-like model ($20k-750k ACV) and initially targeting oil companies to monitor their wells and pipelines. Clients use our technology because we can deploy in 1 week what a legacy industrial company takes 2 years to deploy (if they can at all); this is particularly valuable in the developing world.
We've grown from $0-500k in revenue in the last 6 months; at this point we need to build more technology so we can scale while maintaining reliability.
Very profitable business model in a large market without much competition; we're essentially converting expensive industrial equipment into software with lightweight hardware.
The dataset generated by our sensors is valuable and interesting. We can determine when fuel is being stolen out of pipelines, when there are oil leaks 10,000ft under the ocean surface, and track exactly how much fresh water companies are pulling out of the earth. If we're successful, we'll have an unprecedented level of understanding about the world's fluid infrastructure.
Great investors (YC W16, Matt Krisiloff on the board)
As a small team (3 engineers) we don’t have many formal processes but we do keep a well-ordered backlog of future tasks. We have a continuous deployment system, and generally deploy to production several times per day. New code is reviewed using a GitHub PR (generally within a few hours). We have a pragmatic engineering philosophy. We write automated tests when helpful (which run in our CI pipeline) but our test coverage is currently fairly low. We tend to use trusted (rather than cutting edge) tools and do whatever it takes to ship features fast. That said, we do periodically set aside time to refactor when technical debt is slowing us down. Every engineer is expected to help design an architecture that will support us as we grow the product and the team.
We are trying to turn mountains of ultrasonic data into measurements that accurately describe fluid movement inside of pipes (i.e. flowrates). For this we use a combination of signal processing, physics, and machine learning. There are several challenges that we fight against:
- limited training data about ground source truth in the field (we can generate a lot of lab data, and there are other facets of the system that we break into action/response datasets)
- low-bandwidth and long latency transmissions when we communicate via satellite
- this growing fleet of devices must be managed to a high level of reliability
tracking/prediction of measurement confidence scoring; this will allow us to catch any changes in signal quality (thus accuracy) more quickly
updated installation mobile (web) app; our existing version does not automatically trigger any diagnostics or signal evaluation (ie backend deployment processes are still manual). there is therefore considerable lag before the app can instruct the customer to make physical adjustments during installation.
ml system for training signal quality boundaries on reference measurements (we update this manually now)
Adventurous, self-honest, correctness > pride, brute force when in doubt, humorous
(jerks need not apply)
Our SF office is on 16th and S Van Ness — we think office environment is really important to enjoying work so we put a lot of effort into making it great. You should at least come by and visit, if only just to see the office.
We provide full medical and dental insurance—we even provide life insurance (we're not usually morbid, we just got an absurdly deal on the plan).
We think people should work whenever they feel most productive.
We allow a 3-month fully covered maternal/paternal leave
Interested in this company?
Skip straight to final-round interviews by applying through Triplebyte.