Software Engineer, AI Engineer (Applied​/Software), Backend Developer

36 Labs is an early stage startup using AI and machine learning to optimise the world of social media marketing. We build with video and language models, from how content is conceived, to how it's generated, distributed, and optimised.

We're looking for a Software Engineer to help us build 36 Labs for the long term. Our product moves fast and the codebase has grown fast with it. AI writes a lot of what we ship, so the bottleneck isn't typing, it's judgment and taste. We need someone who can harden the foundations and keep the bar high so we keep shipping at speed as we scale.

• Building and extending the systems that turn video and language models into content, end to end
• Designing the abstractions the team builds on, so the codebase stays an asset, not a liability, as we scale
• Looking for opportunities to reduce complexity in a fast-moving codebase: refactoring, deleting, and making the right thing the easy thing to do
• Integrating and structuring raw platform data into clean, reliable inputs: the formats our downstream systems and AI workflows depend on
• Setting the engineering standards a small team compounds on: testing, observability, CI, and code review

• Strong fundamentals in Python and Type Script, and a track record of writing clean, well-tested production code in large codebases
• Production experience building LLM-powered systems: prompting, context engineering, RAG, agent architectures, evals
• Strong system design skills and product sense, with comfort working across SQL databases and datasets at scale
• Proficiency with AI coding tools: you get more out of them than most
• Can showcase solving hard problems with artifacts like personal projects or Git Hub contributions
• Comfortable reaching into unfamiliar code across the stack to ship the right fix, and working where priorities are a moving target
• Comfortable moving fast where things are loosely defined and priorities compete; you wear several hats and own problems end-to-end