Facet is building artist-centric applications that bridge the gap between tool and assistant, freeing artists to explore new ideas and produce better work. While media becomes increasingly personalized and topical, the process of creating art remains frustratingly manual. We use machine intelligence to amplify human creativity, allowing artists, designers, and creatives of every stripe to realize their visions simply and easily.
We have just launched our first product, a web-based ML-powered image editor, into private beta, and we're expanding our founding team in San Francisco. Facet is backed by Slow Ventures, Basis Set Ventures, South Park Commons, and a diverse group of designers, product thinkers, as well as AI engineers from Google Brain and Salesforce.
Software Engineer (Backend / ML) San Francisco, CA, United States
Software Engineer (Backend) San Francisco, CA, United States
Software Engineer (Frontend) San Francisco, CA, United States
Software Engineer (Fullstack) San Francisco, CA, United States
AI for creative tools is something we at Facet believe very deeply in—we're truly at the cusp of a renaissance in UX design. Tech at the intersection of visual perception and synthesis is already creating to more powerful tools for professionals as well as new forms of media and art (e.g., Cindy Sherman's snapchat-filter pieces). More tellingly, it is creating new ethical challenges, e.g. AI coupled with the rendering and special effects pipelines of Pixar or ILM has deep implications on what constitutes evidence in criminal cases, and has severe potential for abuse for systems like DeepFakes, or that, e.g., can synthesize photo-realistic video of Obama synced up to any audio clip.
Facet is something that has the potential to become huge independent powerhouse, making creative work easier and more accessible for folks of all stripes, and we’re the right team for this problem. We are highly technical and we are backed by AI and product experts from Google Brain, Salesforce Metamind and FAIR. Joe founded two successful companies prior to Facet: Metamarkets was acquired by Snap last year and Premise has expanded to provide basic income and data services to underserved communities in over 30 different countries. Matt has deep expertise in computer graphics and machine learning, and previously led the engineering team at Operator.
As an startup, Facet is high risk, and to be transparent, there is no way we can guarantee our success, but we're well positioned to be smart about it. That said, you’re joining a company at its earliest stage as a member of the founding team, and you’ll have a significant equity stake and influence on our product and culture. We’re going to make mistakes, we’re going to write and rewrite a ton of code, and we’re going to struggle with product direction and market development. We’re being upfront because above all else we value honest communication and collaboration. There is no other way to successfully navigate a seed stage business. By working together, we can go farther, faster, and accomplish more than any of us could by ourselves.
CTO and CEO are both technical, and both currently code > 50% of the time (obviously this will adjust rapidly as the product matures).
Team consists of 2 full-time remote engineers (1 frontend, 1 backend / generalist) and one part-time product designer.
We have a weekly sprint cadence, with daily standups in Slack and issue tracking / discussion on Github. All commits require a code review / second set of eyes. Commits to master run automatically against a testing suite and are deployed continuously to our dev environment.
Real-time browser-based image processing: Performing complex editing operations on millions of pixels in real time isn’t easy. We need to use GPU acceleration, careful memory management, and extensive caching — and we’re looking at any web technology that could give us a boost. We use WebGL extensively, and we’re beginning to use WebAssembly to make sure photo editing always feels fluid.
Splitting state-of-the-art AI models across the browser and the server: Many AI models won’t run in the browser, but we can’t get real-time interactivity by running solely on the server. We are constantly trying to figure out new ways to pre-compute as much as we can on the server, then send compact result summaries to the client that enable sophisticated real-time editing.
Rapidly and dependably deploying new machine learning models: We are constantly training and deploying new models to help us understand photos better. We need flexible and easily-managed infrastructure that will let us change out new models as we improve them, monitor the performance of the models we have, and scale our capabilities to adapt to changing needs.
Building a client-side tool for matching photo colors by example.
Accurate color matching is a core Facet feature that lets users quickly adjust local tone curves and in a content-aware way. Foregrounds, backgrounds, skin tones, etc. are all matched separately and blended together seamlessly. While this operation currently runs entirely on the server, we are porting it to a new Typescript implementation that uses WebGL and Tensorflow.js to run in real-time on the browser.
Training a new object segmentation model for recognizing common photo elements: clothing, skin, sea, sky, ground, etc.
Many commonly-available image segmentation and object detection models are tuned to recognized a large number of different object categories, which can be as specific as individual breeds of dogs. In the photo context, we often care about fuzzier categories, or categories specific to particular customer verticals — for example, a customer that works with fashion photos will be interested in recognizing clothing. We recently trained and deployed two new image segmentation models that recognize clothing and common image background components and provide pixel-accurate segmentations.
Deploying our machine learning models as individual services.
In order to dynamically scale to meet user load, we will need to partition our content understanding pipeline into smaller services, each of which provides a single model, and build the distributed infrastructure, model serving and monitoring to allow them to be scaled and deployed independently.
Consensus goals, independent execution. We value collaboration on goal setting and technical/company direction while giving individuals latitude and ownership over how these goals are achieved.
Decathlon not a sprint. Creative AI is a rapidly blooming space, so we need to move quickly to build product and grow our market. We all juggle multiple roles and we’re in it for the long haul. At the same time, we need to build a stable foundation for our future work, and ensure that we are building thoughtfully, creating a product that makes the world better, and only breaking things if we’re pretty sure we can fix them again. Furthermore, we understand downtime is important. We don’t have an explicit vacation policy, but unlike other startups we expect you to gauge your level of stress and take one when you need it.
Creative+Technical. Facet is about making technology a partner in the creative process. We enjoy product design, engineering, and research—but also photography, music, art, and cooking.
Diverse ideas. We’re solving a completely new problem and need a diverse set of viewpoints and voices to fully understand its scope and extent. We’re building a company and culture that can responsibly and respectfully integrate a plurality of voices.
Tea and baked goods. We’re really into tea and baking & think these are good things. You might disagree with us (cf. “diverse ideas”) but jfyi.
Interested in this company?
Skip straight to final-round interviews by applying through Triplebyte.