Taktile is a technology company building cloud-native software that underpins the financial products of the future. Taktile exists to create value for organizations through smarter and safer decisions at scale. Our goal is to become the world's leading software provider for automated decision-making in the financial services industry and, to date, our software has been used by our customers to power over 15 million credit decisions.
Taktile is based in Berlin and New York City. It was founded by machine learning and data science veterans with extensive experience building and running production ML in financial services. Our team consists of engineers, entrepreneurs, and researchers with a diverse set of backgrounds. Some of us attended top universities such as Harvard, Oxford, and ETH Zurich and some of us have no degree at all. We have accumulated extensive work experience at leading tech companies, startups, and the enterprise software sphere. Our backers include Y Combinator, Index Ventures, firstminute Capital, and stellar angels such as the founders of Looker, GitHub, Mulesoft, Datadog and UiPath.
What You'll Do
- Build a large-scale machine vision system (object detection, image classification, and semantic segmentation) on Google Cloud Infrastructure.
- The system will be used to analyse a large corpus of (already annotated) high resolution aerial images in an insurance context. For example, the system will be able to detect the presence of trees or swimming pools around houses and measure the size of solar panels.
- Benchmark different methodologies and hardware configurations against the current baseline on accuracy, training time, and cost.
- Find a way of increasing the ML system's data efficiency by using a large additional corpus of unlabelled instances.
- Track and present results at regular intervals to internal and external stakeholders.
- Create a codebase that honors software engineering best practices.
- Learn together with your co-workers, by following your intellectual curiosity to discover more about new methods and technologies as well as by participating in our active feedback culture.
- Come into the office (in Berlin) at least a couple of days a week, once the COVID-19 pandemic lets up.
Ideal, But Not Required
- Experience implementing self-supervised learning approaches for machine vision applications from scratch.
- Experience setting up distributed training jobs (on TPUs) with Tensorflow.
- Experience with Google Cloud AI infrastructure.
- Work with colleagues that lift you up, challenge you, celebrate you and help you grow. We come from many different backgrounds, but what we have in common is the desire to operate at the very top of our fields. If you are similarly capable, caring, and driven, you'll find yourself at home here.
- Make an impact and meaningfully shape an early-stage company.
- Experience a truly flat hierarchy and communicate directly with founding team members. Having an opinion and voicing your ideas is not only welcome but encouraged, especially when they challenge the status quo.