Senior AI Research Engineer

At Plumerai, we make it easy and affordable for developers to add highly accurate AI to their embedded devices and thereby enable them to create amazing new products. We combine our on-device Tiny AI software with our cloud-based multimodal LLMs, providing People Detection, Video Search, Familiar Face Identification, AI Captions and more.

Major enterprises deploy our advanced computer vision models on millions of smart home cameras in the field and we're rapidly expanding into commercial security, retail, assisted living, and more. The best solution runs as much as possible on-device to enable low-power, accurate, and private AI products. This is where Plumerai leads, demonstrating better accuracy even than Google Nest. We are now starting to invest heavily in market adoption.

We build the most accurate and efficient AI solutions by vertically integrating all layers of the stack. From data collection and curation, custom training software, model architectures, multimodal LLMs, pre- and postprocessing, and all the way down to the fastest inference engines. Not only does our team have a deep theoretical understanding about all these components, we also know how to ship fast and often.

Our team is based in London and Amsterdam, we have recently raised funding to provide multiple years of runway, while our recurring revenue is growing rapidly. We are backed by world-class investors such as Tony Fadell (creator of iPod, iPhone; founder of Nest), Hermann Hauser (founder of Arm), Zoubin Ghahramani (Google DeepMind), and others.
Learn more here:
TechCrunch: Tony Fadell-backed Plumerai brings on-device AI to home security cameras
Plumerai raises $8.7M Series A to connect Vision LLMs to trillions of edge devices.

Role description

We are looking for a Senior AI Research Engineer that can help us develop state of the art AI products. This can involve anything from improving our training algorithms, training and integrating multimodal LLMs, building our data pipeline, designing new model architectures to using tried and tested ML approaches and coming up with clever algorithms. You will help us build new AI features that will be shipped to millions of camera devices in the field. Together we are building the most advanced AI for embedded devices.

What you will be doing

• We combine our Tiny AI with multimodal LLMs to enable our advanced AI features for our customers. You will use and improve multimodal LLMs to achieve new functionality for our customers and optimize their deployments (cloud and edge).
• Some of our deep learning models are truly tiny - the memory footprint of our smallest computer vision model is just 1MB. You will train and design more accurate models, while also enabling new and more complex AI applications on low-cost and low-power hardware.
• You will improve our data pipeline, model architectures and training software. Sometimes there is relevant literature available, but novel approaches and clever hacks are often required for the problems that we are working on.
• You will use our Kubernetes cluster to deploy PyTorch and TensorFlow training jobs, Snowflake and Dataflow to build datasets, tools like Streamlit to prototype new demos (try one of our live demos here) and lots of GPUs on GCP for training new models and auto-labeling data.

Requirements

What You Need

• +5 years of professional software engineering experience with proficiency in Python.
• Comfortable with frameworks such as PyTorch, TensorFlow, Keras, or JAX.
• Strong experience with computer vision and multimodal LLMs.
• Trained neural networks that moved into production.

Nice To Have

• Industry experience with efficient inference deployments (cloud or edge).
• Experience with Deep Reinforcement Learning.

Benefits

What we offer

• Competitive salary.
• Generous equity stake in the company.
• Relocation assistance.
• Choose your own laptop.
• 25 days of paid vacation time in addition to bank holidays.
• Ability to attend top research conferences like NeurIPS, ICML and CVPR.

We only consider applicants who are currently based in, or willing to relocate to, London or Amsterdam. We have flexible working hours and work together from our offices on at least 2 fixed days per week.