About the job
As a wholly owned subsidiary of the Government Technology Agency (GovTech), Assurity Trusted Solutions (ATS) has established itself as a leading Trusted Partner over the past decade. We provide an extensive range of products and services, including infrastructure and operational services, governance and assurance services, along with managed processes. In today's fast-evolving digital and cyber landscape, where trust and collaboration are paramount, ATS is dedicated to fostering mutually advantageous business outcomes by working closely with GovTech, government agencies, and commercial partners to mitigate cyber risks and enhance security frameworks.
We are seeking a skilled AI Engineer to develop agentic AI systems for various cybersecurity applications. This role integrates large language models (LLMs) with robust AI/ML foundations, including data pipelines, classical machine learning where applicable, thorough evaluations, and essential safety measures. The position is a direct contract with us until March 31, 2027, with the possibility of extension based on performance.
Key Responsibilities:
- Design, develop, and deploy agentic AI functionalities such as planning/execution loops, tool utilization/function calling, and multi-step workflows aimed at security applications like vulnerability triage, assistive exploit reproduction, and incident-response support.
- Implement and strengthen retrieval-augmented generation (RAG) techniques, encompassing indexing, chunking, routing, re-ranking, feedback loops, and data governance tailored for sensitive environments.
- Establish evaluation and observability frameworks for LLM/agent workflows, including tracing, cost/latency/quality dashboards, and both offline and online evaluations to facilitate product enhancements.
- Develop safety measures and guardrails including content policies, schema/output validation, PII redaction, prompt-injection defenses, and controlled tool permissions, while monitoring these mechanisms in production environments.
- Utilize traditional machine learning techniques for tasks like classification, regression, and anomaly detection when advantageous; conduct A/B testing and error analysis to identify the optimal approach.
- Take ownership of productionization processes, including CI/CD for AI applications, containerization, scalable inference endpoints, vector/search infrastructure, and operational runbooks with established service-level objectives (SLOs) for reliability.
- Collaborate effectively with product and security teams to define challenges, produce concise design documentation, and rapidly iterate while adhering to security and privacy standards.
- Carry out other assigned tasks as necessary; responsibilities may evolve with the product requirements.

