Sparta Commodities
Remote, Spain · Remote · Full-time
Loading job details
Preparing the role information
We Are Sparta Sparta is the AI leader for commodity trading, home to Leonidas AI, the first operating system for oil and commodity traders. Backed by $42 million in Series B funding led by One Peak, with participation from returning investors FirstMark and Singular, the same investors behind AI leaders like Legora, Dataiku, and Synthesia. Most companies are building a chatbot on top of data. We've built an agent with an actual reasoning layer, grounded in years of proprietary market data and in-house trading expertise nobody else has. We're scaling fast, and it's an exciting time to join. From independents to multinationals, the world's leading commodity teams use Sparta and Leonidas AI to move faster, trade smarter, and stay ahead of the market. At Sparta, you'll be trusted to take ownership, backed by a team who wants you to succeed. AI is how we build here, not just what we sell. Good ideas move fast, and yours could go from staging to production in a couple of hours, whatever department you're in. You'll be challenged, supported, and given room to grow, building the AI that's changing how commodity trading gets done. Your work has reach. From shaping Leonidas AI, to powering smarter decisions for traders around the world, to changing how commodity trading is done. And because we're growing fast, you'll accelerate your career with a level of ownership and impact you simply won't get at bigger, slower-moving companies. The role You are the senior technical authority for how Sparta builds agents. You design and build the harness that runs Leonidas AI's agents, set the patterns the rest of the team and the wider engineering group follow, and stay in the code on the hardest problems. You work as a peer of the Agents Team Lead: they own the team and the delivery, you own the technical direction of the agent platform. When a trader asks why an arbitrage is shut, Leonidas has to pick the right curves, call the right tools, reason across them and show where every number came from, in seconds. Models, frameworks and best practice for agents change every few months. Your job is to build a harness that absorbs that change without rewrites, and to make sure we can prove each change made the product better. What you will do Set the technical direction. Author the target architecture for the agent platform: execution graph, LLM interfaces, tool contracts, memory and context management, sub-agents, sandboxing and how skills plug in. Write the designs, run the reviews and own the trade-offs. Hands on AI development - Context engineering across branched conversations, handoffs between specialist agents, a decision-model layer so a model can change without touching agent code, and resumable, observable runs. Make quality measurable. Define how we evaluate agents: golden question sets, offline and online evals in CI, tracing and feedback loops, and release gates a model or prompt change must pass. Open Leonidas up. Design the agent-to-agent API, MCP server and headless modes, and the boundaries between what we expose, what customers connect and what stays internal. Own reliability, cost and safety. Latency and token budgets, failure handling, guardrails, provenance on every number, and the security model for agents that call tools and run code. Raise the bar for everyone. Mentor engineers, coach the team through design and code review, and spread good agent-building practice, including AI-assisted development, across all of engineering. Manage and grow people. Line-manage a group of engineers on the Agents team: run 1:1s, set goals, give feedback, support career growth and take part in hiring. Shape the roadmap. Work with the Team Lead, product and customers to decide what to build, what to buy and what to stop, and to size the risk in new ideas through quick prototypes. What you bring 8+ years building production software, including designing and shipping LLM-powered agents or agentic products used by real users. Technical leadership without the title: you have set architecture across a team, led designs through review and mentored others to a higher level. Deep, hands-on knowledge of agent systems: tool calling, multi-agent orchestration, streaming, context and memory management, retrieval, and failure modes over long runs. Clear writing and influence: you explain trade-offs to engineers, product and executives, and you change minds with evidence. Nice to have Built an agent harness or framework, or contributed to an open-source one (OpenAI Agents SDK, Claude Agent SDK, LangGraph, DSPy or similar); built MCP servers or clients. Knowledge graphs, ontologies or entity resolution (mapping trader language to the right data). Sandboxed code execution, guardrails or security work for agents. Fine-tuning or model selection experience, or a research background. Interest in commodities or financial markets. Our stack for this team Python · OpenAI Agents SDK · Postgres · Redis · StarRocks · Kafka · AWS · Nx monorepo · GitHub Why J
Sparta Commodities
Posted via Ashby_ats
Apply Now opens Rozgoo's sign-up and application flow for this role. Updated today.
Apply Now