Entrupy aims to establish itself as the global standard for authenticating physical goods. We enable businesses and consumers to instantly verify the authenticity of high-valued physical objects, starting with luxury goods (e.g. leather goods and accessories). Our mission is to protect businesses and consumers from transacting in counterfeit goods. Currently in use by hundreds of secondary resellers and marketplaces worldwide, Entrupy provides the only scalable technology capable of authenticating high-value products.
Our yearly revenue has increased by 140%, customer growth by 50% and authentication volumes by 150%. We are scaling up and experiencing a lot of growth.
Two of our investors are quite well known: Yann LeCun (VP and Chief AI Scientist at Facebook, ACM Turing award), Zach Coelius (early investor in Cruise Automation which sold to GM for $1B)
We still have one foot in academia and always work on cutting-edge machine learning and computer vision.
Being both a hardware and software company, you get to experience the entire hardware/software life-cycle. Also, the technical challenges are very unique.
Our engineering team is based in New York and has five members who work on mobile apps, machine learning, systems and internal tools. We also have three remote members who work on hardware and ML research. Engineers take ownership of projects and system areas and collaborate across the entire company. As an early engineer on a small team, you'll have the opportunity to work on a wide variety of challenges and projects and define important pieces of a product that can't be found anywhere else.
Entrupy's technology enables authentication of an ever-expanding catalog of luxury goods. Users take microscopic images of different areas for each item with our mobile app and lens attachment. Machine learning systems run a battery of in-depth checks, assessing the relevant properties of each item to determine its authenticity.
Core areas include the following:
Machine learning: Our core product is powered by ML systems capable of authenticating a massive variety of products. Doing this requires deep analysis and research, and iterative development of targeted tests that deeply understand each item. These systems need to be robust to real-world use and misuse by users across a range of industries and contexts, and their output needs to be comprehensible to a non-technical audience.
Mobile apps: We have a wide range of customers from small pawn shops and resellers to high-volume enterprises. Our apps need to cater to different workflows and be intuitive to users with limited knowledge of luxury goods. Real-time capture guidance and integrated support communication are important pieces to create a high quality experience.
Platforms and infrastructure: Our ML products are supported by internal pipelines and platforms capable of handling both live production requests and training fully auditable, complex models from diverse data sources. Our Vue.js dashboard platform and underlying real-time APIs power support, training, knowledge and analysis tools. Heavy automation and careful service design allow a small team to support many use cases.
Scale up machine learning products. How responsive can models and systems be to both increases in data for a single product and additions to product lines? What's the best way to quickly get improvements from research into the hands of users? How can we continuously monitor and improve systems in the wild?
Improve support tooling and knowledge platforms. What's the most relevant information to describe an item? How can annotation be framed so that other teams and automated systems share understanding?
Build high-availability API and data platforms that can scale and serve a variety of use cases. What types of systems balance short-term needs with longer term predictions?
Figure out common reasons for items to require manual review, then work cross-department to discover and implement solutions. Many problems can be approached from client, backend, ML, support and customer training angles: what's the best way to drive improvements holistically?
Understand the boundary between variation due to user behavior, hardware variance, usage patterns and differing real-world objects. What are the different ways people can use the system? The range of counterfeits is essentially infinite: what should be tested for?
- We care about the problem and the mission.
- We believe in results. We don't care where/how you work.
- Being nice to others is a must. But be vocal about opinions.
- Flat structure.
- Responsibility and autonomy.
Interested in this company?
Skip straight to final-round interviews by applying through Triplebyte.