Co-founder & CTO; Stealth AI Infrastructure Physical Commodity Trading Venture

What we do

Physical commodity trading is one of the world’s largest and least digitised industries. Critical commercial workflows still run through messages, calls, spreadsheets and tacit trader knowledge, while incumbent systems are generally better at recording transactions than preserving the full decision chain behind them. Our thesis is simple: a trader is an information processor, and human bandwidth is the bottleneck
. Physical commodity opportunities develop over hours or days, not milliseconds. The edge comes from broader coverage, better information processing and reducing the time from fragmented market signal to a priced, risk-checked opportunity. We are building an AI-native operating layer for physical commodity trading
, developed and proven through an integrated trading operation. The system is designed to help a small commercial team process materially more information, evaluate more opportunities and execute physical flows with greater speed and consistency. We begin with a small number of repeatable physical flows selected for commercial readiness, capital efficiency and technology leverage, then expand across adjacent industrial commodity markets.

Over time, selected external flow originators can operate through the same infrastructure, allowing commercial reach to scale beyond internal trading headcount. The company is founded by a physical commodity trader with 15+ years across leading global trading houses. It is being built with Merantix Capital, a leading European AI venture studio, with working-capital structures and initial commercial flows already under development.

The trading operating system is the core technology of the company. It is built around live commercial workflows rather than as generic software for incumbent trading houses and may later support selected external flow owners through controlled operating and revenue-sharing models. You will design and build it from first principles:

• The trading cockpit. A single workspace that turns fragmented market and commercial information into structured, actionable intelligence.
• A proprietary intelligence platform.A proprietary representation of counter parties, relationships, flows, market context and trading outcomes.
• An agentic layer on top. Systems that support opportunity identification, commercial analysis, pricing, risk and execution while keeping the human in control.
• A learning loop. Every signal, decision and settled trade enriches the system and improves future decision support.
• Workflow automation. Selective automation of high-value operational processes.

• Ship the first product yourself. You are a hands-on builder who can take the platform from zero to live trading support solo, before hiring a team.
• Own the full technical stack and architecture: data ingestion, ML/agent layer, and the trader-facing product.
• Co-define the company strategy with the co-founder & CEO: the 10 year vision, the platform strategy, and which verticals and flows we scale into first.
• Go into the field. Sit with traders, visit counter parties, and translate how physical trading actually works into systems that map to trader P&L logic and workflows.
• Pitch investors alongside the CEO and own the technical narrative through Seed stage and beyond.
• Recruit and lead the founding engineering team as activity scales.

• A strong mathematical, machine learning, and data engineering foundation.
• Applied AI/ML and data architecture depth: you have built and shipped real AI products into production, in messy, real-world environments.
• Experience turning heterogeneous, unstructured, multi-source data into decision systems: knowledge graphs, agentic systems, LLM orchestration, or signal/time-series pipelines.…