Senior Python Engineer: AI Agents & Forecasting
Upload My Resume
Drop here or click to browse · PDF, DOCX, DOC, RTF, TXT
Requirements
• 5+ years building production-grade Python backends. You know the internals, not just the syntax. • Hands-on experience with LLM orchestration frameworks such as LangChain or LangGraph (agent memory, tool calling, state management) • Dagster in production (assets, sensors, partitions) or equivalent pipeline orchestration • Strong backend fundamentals including API design, async programming, and database modelling • You've built systems that had to work reliably at scale, not just pass a demo • 5+ years in a backend or full-stack engineering role • Worked at an early-stage startup or high-growth environment. You understand the pace. • Built and shipped production systems, not just prototypes • Comfortable being the most senior engineer in the room, or the only one • What Sets You Apart • You think holistically about systems. You see how your work connects to every other part of the product without being told. • Research-driven approach to problem solving. You test hypotheses, not just ship features. • Experience in financial markets, algo-trading, or prediction market platforms (Polymarket, Manifold, etc.) • Quantitative background in maths, statistics, or probability theory • ML experience including random forests, regression models, scikit-learn, and PyTorch • AWS infrastructure experience deploying containerised applications and managing cloud environments (EC2, Lambda, RDS) • Open-source contributions to AI or crypto projects
Responsibilities
• Unifying the different parts of the stack. The network, signal layers, and forecasting architectures need to work together as one coherent system. • Designing and building the agent orchestration pipeline that allows AI forecasters to ingest data, reason, and produce predictions. • Building and optimising the signal pipeline that feeds real-world data into forecasting models. • Experimenting with different forecasting architectures to find optimal approaches for linking targets to signals. • Writing production-grade Python that handles complexity at scale, not scripts that work in a notebook. • Contributing to technical strategy alongside the founders. You'll have a voice in what gets built and why. • Evaluating and vetting technical candidates as the engineering team grows. • We're early. The engineering function is still being built. • Documentation is thin and processes are light • Priorities will shift and things will break • If you need a mature engineering culture with established practices, this isn't the role yet. You'll be helping build that culture, not inheriting it. • More ownership, influence, and impact in 6 months than you'd get in 5 years at a bigger company • Technical co-founders who are direct, move fast, and actually understand what you're building • No politics, no layers, no waiting for approval • The problem is genuinely unsolved and the market is enormous • If we get this right, you'll have built the infrastructure behind the world's most advanced forecasting AI. • Make the Numinous platform into a self-improving world model with the most accurate forecasts in the world. You exist to turn our vision of self-improving superhuman AI forecasting into a reality, building the systems that allow agents to get smarter, faster, and more accurate over time. • What Good Looks Like • At 3 months: There's a production-grade forecaster in one category. Targets are linked to signals, connected to the network, with experimentation underway on forecasting architectures. You're making architectural decisions independently and contributing to technical strategy, not just executing tickets. • At 6 months: The signal pipeline has an inductive process, a world model sits on top, and there's a sellable product for clients. You've unified the different parts of the stack (the network, signal layers, and forecasting architectures) into something coherent and powerful. • At 12 months: The platform handles complex targets across multiple verticals including commodities, politics, and beyond. There's a provable edge that unlocks serious commercial conversations. You're a technical leader in the company, not just an engineer.
Benefits
• This is founding-engineer territory. The team is small, the problem is unsolved, and the market is massive. You're not joining a machine that's already running. You're building it. The architecture decisions you make in the next 6 months will define the platform for years. You'll have direct influence over the product, the technical direction, and the culture of the engineering team as it grows. • In 12 months this role won't look like this. The team will be bigger, the foundations will be set, and the opportunity to shape everything from scratch will be gone. Right now, it's all open. • What Good Looks Like • At 3 months: There's a production-grade forecaster in one category. Targets are linked to signals, connected to the network, with experimentation underway on forecasting architectures. You're making architectural decisions independently and contributing to technical strategy, not just executing tickets. • At 3 months: • At 6 months: The signal pipeline has an inductive process, a world model sits on top, and there's a sellable product for clients. You've unified the different parts of the stack (the network, signal layers, and forecasting architectures) into something coherent and powerful. • At 6 months: • At 12 months: The platform handles complex targets across multiple verticals including commodities, politics, and beyond. There's a provable edge that unlocks serious commercial conversations. You're a technical leader in the company, not just an engineer. • At 12 months: • Salary: £80,000 to £120,000 depending on experience • Equity: Competitive for the right candidate • Hours & Location • Remote-friendly with a presence in London • Async by default, in-person when it matters • We care about output, not hours logged. This is a startup, and the pace reflects that.