About Pendulum

Pendulum is an immersive language learning company. Language is the instrument of culture. When we learn to speak a new language, we learn to see a bigger world. ****But the way we learn new languages is broken. We don’t learn to speak by memorizing vocabulary, drilling conjugations lists or playing daily games with a green owl. We learn by immersing ourselves in the language—by hearing it, seeing it, and using it as a vessel for connection. Pendulum brings this to life for the first time ever in a digital format by leveraging the latest in conversational voice AI and autonomous LLM agents. We create immersive, personalised and ever-evolving digital worlds for language learning where users step into dynamic narratives and conversational roleplay tailored to their interests, personality, and fluency level. Instead of rote grammar drills or generic dialogues, learners engage in evolving, open-ended scenarios where they actively shape their journey through language.

Role

As an AI engineer at at Pendulum, you will own and scale the backend systems that power our voice-first, AI-driven language learning experience. Your role will be pivotal in developing the infrastructure that enables real-time, multimodal AI interactions. Examples of recent AI engineering work we have shipped include: giving our core conversational LLM agentic abilities to autonomously generate images based on conversational context; implementing memory in our LLM system to make a more personalised language companion, and creating an agentic system to create a ‘Daily Review’ feature that draws from a user’s entire conversational history.

You will:

• Design and optimise systems that orchestrate multiple LLMs, voice AI components (TTS, STT), and visual AI to power real-time conversational language learning backed by voice interactions and visual support. This will involve handling streaming responses, profiling and optimising components of chained AI systems, memory management, and agentic workflows.
• Own and scale our FastAPI-based backend, ensuring performance and reliability as we grow to hundreds of thousands of users.
• Develop and scale WebRTC-powered voice interactions, leveraging and modifying open-source frameworks like Pipecat to fit our use cases.
• Ensure a smooth developer experience by writing and maintaining clean, well-documented code supported by sensible system design choices.

Qualifications

• 5+ years of backend engineering experience, with strong expertise in Python. Deep experience with FastAPI or similar Python web frameworks is a must.
• Experience building and scaling LLM-powered applications, particularly those involving visual AI, agentic workflows, or voice-based AI interactions.
• Understanding of asynchronous processing and distributed task execution.
• Familiarity with cloud infrastructure and serverless deployments, preferably on platforms like Modal.
• Strong software engineering experience with the ability to pragmatically think through software design and architecture to ship rapidly while scaling to hundreds of thousands of users.
• You can ship at high velocity without compromising on quality and are comfortable figuring out how to build things you haven't built before.
• Experience working with WebRTC or real-time media streaming technologies is desirable.

Our Stack

• Frontend: React with Next.js, deployed on Vercel.