We're looking for an Architect of Autonomous Research Platform to architect and build, from the ground up, the agent-based research and execution pipeline for new HFT initiative. You will not be integrating off-the-shelf LLM APIs. You will design a multi-agent system where specialized agents collaborate.
The Project:
You’ll be working in a vertically integrated setup where AI research is directly translated into trading strategies and real-world performance.
The long-term vision a fully automated pipeline where the entire lifecycle: hypothesis → research → validation → execution → optimization is handled by agent-based systems.
Responsibilities:
Designs and implements a multi-agent orchestration framework for HFT research, defining agent roles, communication protocols, memory/context management, and iterative decision loops
Builds retrieval and knowledge systems that enable agents to ground reasoning in market data, research papers, and internal artifacts
Develops a Strategy DSL and compiler for safe, structured representation of strategies across research, simulation, and live execution
Implements a falsification-first evaluation layer to filter false alpha (overfitting, lookahead bias, non-robust signals) prior to capital deployment
Creates an alpha memory system to track hypotheses and outcomes, preventing repeated exploration of invalid ideas and enabling cumulative learning
Collaborates with HFT researchers and engineers to ensure seamless transition from agent-generated ideas to low-latency production systems
Defines robustness metrics beyond P&L, including regime stability, capacity, execution realism, and novelty
Required Skills & Experience:
LLM architecture expertise: RAG, fine-tuning, prompt engineering, and evaluation frameworks
Agent systems experience: Building multi-agent orchestration, memory management, tool use, and collaboration (beyond basic LLM integrations)
Experience creating auto-researcher / co-scientist systems: Proven track record of building autonomous research agents or AI systems that assist scientists/analysts.
Strong Python + ML skills: Production-ready code with PyTorch, JAX, or similar frameworks
Statistical rigor: Experimental design and statistics for non-stationary, noisy environments
Systems thinking: Ability to design abstractions, interfaces, and pipelines—not just models
English proficiency: B2+
Nice to Have:
HFT/MFT Experience & Low-Latency Mindset:Prior experience in High-Frequency/Market-Making Trading or working with low-latency constraints (nanosecond-scale, hardware-aware, deterministic design) is a plus.
What’s in it for you?
Compensation is competitive and will be discussed based on your expectations and experience.
Location: Montreal (Canada), the Netherlands. Official employment or Remote work under a B2B contract from the European time zone.
Relocation support - available for the Netherlands.
Access to an exclusive computing architecture (an alternative to NVIDIA GPUs), for which the team will have early access.
Involvement in a research-driven project with industrial applications, rather than a standard production environment.
Work within a young, agile team with no bureaucracy, where decisions are made quickly and efficiently.