Founding AI Engineer, Pendulum
£100-140k
+ Significant Equity
A new kind of language learning company
Job no longer available
A new kind of language learning company
1-20 employees
Job no longer available
£100-140k
+ Significant Equity
1-20 employees
Company mission
Our mission is to open access to the world's culture through democratising language learning
Role
Who you are
- You have a builder mindset. You have a track record of shipping products in a fast-paced environment
- A bias to productionise. You want (and know how) to push cutting edge AI to customers, in order to solve real-world problems
- You thrive in a startup environment. You’re excited to solve user problems through quick iterations, you’re ok with priorities changing and you’re comfortable with risk
- Proficiency in Python and a deep understanding of PyTorch
- Familiarity with open source generative models and workflows for image generation, such as Stable Diffusion, ControlNets, IP-Adapters, LoRAs, LCMs etc. and the ability to make them work in concert
- Experience with preparing and processing training datasets from unstructured raw data
- Experience with model fine-tuning and personalisation, such as with LoRAs, Dreambooth, Textual Inversion etc.
- Strong software engineering experience with software design / architectural skills
Desirable
- Experience as an early engineer at a fast growing startup
- Understanding of common architectures in generative AI and their implementations (e.g. Latent Diffusion Models, Diffusion Transformers, Attention, VAEs.)
- Experience with Huggingface libraries such as diffusers for image and/or video generation
- Familiarity with image generation workflow technologies such as ComfyUI, Fooocus or Automatic1111
What the job involves
- Experiment with, fine-tune, and orchestrate image generation models to power controllable and high-quality generation across a variety of specific workflows
- Engineer pipelines to generate structured datasets for automated model fine-tuning (all our data is fully licensed from creators and brands we partner with)
- Deploy and optimise models to scale inference for thousands of users across our partners’ platforms
- Ensure the reliability and efficiency of our systems across the entire ML stack, from data collection, experiment tracking, model training and beyond
Application process
- 1. Apply via Otta
- 2. Video call with CTO
- 3. Short take-home exercise involving our own product and code base
- 4. Video call with the whole team
- 5. Offer
Salary benchmarks
Our take
Pendulum is an immersive language learning company, developing AI products that redefine the very basics of learning a new language. Pendulum develops its products off the philosophy that to truly learn a language, one must experience the culture, differentiating it from the gamified learning of Duolingo and the traditional focus of education on grammar and sentence memorising.
Its products leverage conversational voice AI and autonomous LLM agents to create evolving digital landscapes that are fluid to the customer's fluency level. Essentially, customers are immersed in open-ended scenarios that are actively shaped by their conversations, thus replicating the experience of speaking to someone in a foreign language
Founded by a small team emerging from the South Park Commons Founder Fellowship, Pendulum has garnered support from notable investors, including Shine Capital and Mercuri, and early backers from platforms like Notion, Vimeo, and Patreon. These endorsements reflect confidence in Pendulum's approach to learning new languages. The team behind Pendulum brings extensive experience in AI product development, having previously built solutions at Spotify, led product and strategy initiatives at Google, and conducted research on generative media and natural language processes at Oxford and Imperial College. With this extensive background, Pendulum is ready to transform the way we learn new languages.
Kirsty
Company Specialist at Welcome to the Jungle
Company
Funding (1 round)
Jun 2024
$3.8m
Seed
Total funding: $3.8m
Company HQ
London, UK
Leadership
William Usdin
(Co-Founder)Previously worked at Google through several roles, most recently as Head of Strategy and Operations, EMEA. Also founded Woven Science and acted as their Chief Operating Officer.
Zo Ahmed
(Co-Founder)Ex-Spotify, most recently as their youngest ML Engineering Lead. Built out Spotify's podcast recommendation engine from scratch. Previously researched deep learning at Oxford University.
Share this job