AI Systems Engineer

SIGNALPLUS SG PTE. LTD.; Singapore; On-site; SGD 120,000 - 180,000; Full time

Job summary

SignalPlus SG PTE. LTD. is seeking an AI Systems Engineer (Model Routing) to design and build automated routing systems across multiple foundation models, optimizing task-to-model assignments based on type, complexity, and performance signals.

You will also create evaluation and production infrastructure to measure model capabilities, translate signals into routing decisions, and continuously improve routing performance in production.

Qualifications

• Hands-on experience building production AI, ML, or distributed systems.
• Hands-on experience with automated model routing, dynamic model selection, model arbitration, ensemble systems, or multi-model decision systems.

Responsibilities

• Design and implement automated model-routing and dynamic model-selection systems across multiple models.
• Determine the most appropriate model for each request based on signals such as task type, complexity, model capability, quality, latency, cost, and reliability.
• Develop model-selection, arbitration, escalation, fallback, and ensemble mechanisms for different production scenarios.
• Translate evaluation results into actionable routing policies and continuously improve routing in production.

Skills

Automated model routing

Production AI systems

Multi-model infrastructure

Evaluation & benchmarking

Job description

AI Systems Engineer

We are looking for an AI Systems Engineer (Model Routing) with strong hands-on experience in automated model routing, multi-model systems, AI evaluation, and production AI infrastructure.

我们正在寻找一位 AI Systems Engineer (Model Routing)，候选人需要在自动化模型路由、多模型系统、AI 评估及生产级 AI 基础设施方面具备扎实的实际经验。

Role Overview｜岗位概述

The core responsibility of this role is to design and build systems that can automatically determine which underlying foundation model should handle each incoming task or request, based on factors such as task type, complexity, model capability, quality requirements, latency, cost, and reliability.

该岗位的核心职责是设计并建设能够针对每一个输入任务或请求，自动判断应由哪个底层基础模型处理的系统，并综合考虑任务类型、复杂度、模型能力、质量要求、延迟、成本及可靠性等因素。

You will also build the evaluation and production infrastructure required to measure model capabilities, translate evaluation signals into routing decisions, and continuously improve routing performance in production.

你还将负责建设相应的评估及生产基础设施，对不同模型能力进行量化，并将评估结果转化为模型选择及路由决策，持续优化生产环境中的路由表现。

Key Responsibilities｜岗位职责

Model Routing & Decision Systems｜模型路由与决策系统

Design and implement automated model-routing and dynamic model-selection systems across multiple proprietary and open-source foundation models.

Determine the most appropriate model for each request based on signals such as task type, complexity, model capability, expected quality, latency, inference cost, reliability, and operational constraints.

Develop model-selection, arbitration, escalation, fallback, and ensemble mechanisms for different production scenarios.

Explore and apply approaches such as classification, ranking, learned routing, contextual bandits, or other data-driven decision methods where appropriate.

Continuously improve routing policies using offline evaluation, online experimentation, production feedback, and controlled A/B testing.

Measure routing performance against fixed-model baselines and quantify improvements in quality, latency, cost, and reliability.

This role focuses on model-level routing based on task requirements and model capabilities, rather than simple load balancing, availability-based failover, static configuration switching, or agent/tool routing.

Evaluation & Benchmarking｜评估与基准体系

Define task taxonomies, evaluation dimensions, scoring criteria, and acceptance thresholds required to compare models and support routing decisions.

Design and maintain benchmark suites, golden datasets, annotation standards, and regression test sets.

Build automated evaluation pipelines using deterministic checks, LLM-as-a-Judge, rubric-based scoring, pairwise comparison, and human evaluation where appropriate.

Validate evaluation methods against human judgments and monitor judge consistency, bias, and drift.

Establish continuous evaluation and regression mechanisms for changes to models, prompts, data, and routing policies.

Translate evaluation results into actionable model-selection and routing policies rather than treating evaluation as an isolated benchmarking exercise.

Multi-model & Production AI Infrastructure｜多模型与生产级 AI 基础设施

Design scalable infrastructure for integrating and operating multiple proprietary, open-source, and self-hosted models.

Build model gateways, unified APIs, version-management mechanisms, routing infrastructure, and model lifecycle management capabilities.

Support traffic governance, model rollout, rollback, replacement, and controlled experimentation.

Build monitoring, logging, tracing, failure analysis, and performance diagnostics for both model calls and routing decisions.

Establish feedback loops capturing model performance, routing outcomes, failure cases, user signals, and human-review results.

Build production-grade AI systems with strong standards for reliability, scalability, security, testing, and maintainability.

Qualifications｜任职要求

Hands-on experience building production AI, machine-learning, or distributed systems.

Demonstrated hands-on experience with automated model routing, dynamic model selection, model arbitration, ensemble systems, or multi-model decision systems.

Experience ma