Trading Systems Engineer

About the role

Tokka Labs is a Singapore-based proprietary trading firm and market maker in digital assets and tokenized real-world assets. It is looking for a Trading Systems Engineer to build and scale the low-latency systems powering its high-frequency trading across on-chain and traditional markets, working closely with quantitative traders and researchers. The role is for engineers who have built and run trading systems in production and want to use AI to change how the craft is done.

Key responsibilities

• Design and develop low-latency, high-throughput systems for trade execution, order management, real-time market data and exchange connectivity.
• Optimize performance and ensure operational stability through profiling, tuning and architectural improvements.
• Lead by example in using AI agents to ship faster without compromising reliability, shaping the self-learning loops in the SDLC (evaluations, telemetry and feedback).
• Set the rules for where agents can act on their own and where human review is required, especially for code on the live trading path.
• Work with traders and researchers to turn trading ideas into production-grade systems, bringing AI into how they are researched, tested, deployed and monitored.

Required profile

• A strong track record of building and maintaining low-latency systems for algorithmic trading.
• Experience working directly with traders or researchers on strategies, research or simulation tooling.
• Strong, well-reasoned opinions on how AI should change trading systems engineering, and sound judgment about risk.
• Prior experience in digital-asset markets is not required.

Required skills

• Deep understanding of trading system architecture: exchange and broker connectivity, order management, pricing and risk engines, and smart order routing.
• Strong low-level systems engineering skills, with hands-on experience in performance tuning, monitoring and debugging.
• Genuine interest in AI: using agents seriously in own engineering work and pushing them beyond autocomplete.