Job description
We are hiring a Staff / Principal AI Engineer responsible for designing and building the core agent models of the trading platform. You will lead the complete lifecycle of the model from architecture design, training fine-tuning to inference deployment and real-time monitoring, developing autonomous trading agents capable of analyzing market data, performing multi-step reasoning, and executing strategies under real capital and latency constraints.
Responsibilities
* Design and develop agent trading models to achieve autonomous execution of planning, reasoning, tool invocation, memory, and cross-market data and trading venues.
* Responsible for the full lifecycle of the model, including data pipelines, training and fine-tuning, evaluation, inference services, monitoring, and continuous optimization.
* Optimize the trading reasoning and decision-making capabilities of the model using methods such as Full Fine-tuning, LoRA / PEFT, RLHF / RLAIF, and Reward Modeling.
* Optimize low-latency, high-throughput inference systems, including quantization, batching, caching, distillation, and model service infrastructure.
* Build backtesting and simulation environments to evaluate the decision quality, stability, and risk of agents before deploying real capital.
* Establish guardrails, risk control, and observability systems to ensure agents operate safely, stably, and predictably in production environments.
* Set AI technology standards and guide the AI and quantitative engineering teams.
Requirements
* Over 7 years of production-level ML / AI system development experience, with practical delivery experience across the complete model lifecycle.
* Deep understanding of model fine-tuning, PEFT, reinforcement learning methods, as well as inference latency, throughput, cost, and service optimization.
* Experience in developing agentic AI systems, familiar with planning, tool invocation, memory, multi-step reasoning, and agent orchestration.
* Proficient in Python, familiar with PyTorch, Hugging Face, vLLM, TensorRT, or similar tech stacks.
* Solid software engineering and systems engineering skills, capable of handling large-scale real-time data and time series data.
* Rigorous evaluation-driven mindset, able to validate model performance through backtesting, simulation, and metrics.
Preferred Qualifications
* Experience in trading, quantitative finance, digital assets, or other low-latency, high-risk-sensitive fields.
* Experience in developing reinforcement learning and sequential decision systems.
* Experience deploying and operating models under strict reliability, risk control, or regulatory requirements.