Tractable is hiring a

Deep Learning Engineer

London, United Kingdom


Tractable is a Deep Learning start-up, founded in 2014 and backed by leading investors from the Silicon Valley ($10m funding to date). At Tractable you will be working with a tight-knit team, taking a data-driven approach to solve real-world machine learning problems with deep learning. We have a track record of taking cutting edge research and making it work for real products: we have a team of experienced researchers and engineers, the necessary resources (in-house labelling and domain experts) and a large volume of data (100M+ images) to train our algorithms.



Our main focus at the moment is on the auto insurance industry: we are building models that can understand the ways in and extent to which a car is damaged. As opposed to most applied DL problems, this is a visual task on which trained experts with decades of experience require ~30 seconds of thinking time to arrive at a decision.We have currently modelled most of the external parts of a car, which is the basis of our audit product currently being used by 5+ major insurers. In addition to further improvements on these models we are now turning to combining them with metadata and even more sophisticated visual models, with the objective of further accelerating the process of vehicle damage assessment.



As a Deep Learning Engineer, you will bring research insights from proof-of-concept to a production-quality solution. In doing so, you will have to tackle both research DL questions (figuring out how a solution's performance will scale to 100x more data) and work with all levels of our technology stack (systems to handle TBs of data for machine learning, analytics and real time processing workloads).

Your will work on scaling and productising machine learning solutions at every stage of our product pipeline, from optimising our labelling process to figuring out how to generalise an existing product to new geographies. You will have an intuitive understanding of DL concepts, to an extent that you could explain them to a 9 year old.

For the broader stack, we use Big Data technologies (Spark, Kafka, Cassandra) and state-of-the-art infrastructure and tools (Docker, Mesos) running on a mix of public cloud services (AWS, Azure) and bare metal. Our principal languages are currently Python for model deployment / development and TypeScript for backend.

Other jobs at Tractable