Applied AI Research Engineer Agentic Systems

Nanyang Biologics (NYB) is a biotechnology company leveraging AI to revolutionize drug discovery and develop innovative therapeutics inspired by nature. Through advancing AI-driven solutions in the pharmaceutical industry, we address the complex and costly challenges of discovering effective treatments for difficult diseases. With a global track record of delivering projects with leading companies across the USA, India, Japan, Korea, Singapore, etc., our team of experts based in Vietnam and Singapore works collaboratively across functions and leverages global expertise to drive innovation and excellence.

Responsibilities

Who We Are Looking For

NYB is seeking a senior engineer with genuine research instincts who treats our agent system as a living research problem. The agentic field reinvents its best practices every few months; we need someone who reads the frontier, has the taste to tell signals from hype, prototypes the promising ideas, and lands the ones that work in production - with evidence they made the system better.

You're not here to stand up infrastructure from scratch; we have a working system orchestrating 300+ tools and a small team running it. You're here to make it markedly smarter, more reliable, and more autonomous - and to raise the bar for the engineers building alongside you.

Why This Role

Most agentic AI work optimizes against benchmarks that don't push back. Here, the agent's hypotheses get tested at the bench - its reasoning is answerable to real experimental results, not a leaderboard. We have a working system orchestrating 300+ tools; what we need is someone to keep it at the frontier. You'll read the latest in agentic research, decide what's real, and ship it into a system where "better" means a better experiment runs next week. If you want your work on autonomous agents to be grounded in reality instead of demos, this is the rare seat where it is.

Roles and Responsibilities

• Own the research-to-production loop for our agent: survey emerging agentic techniques, run structured experiments against them, and ship what proves out.
• Continuously upgrade core agent capabilities - planning, tool selection and reasoning, memory, multi-agent coordination - measured against real benchmarks, not demos.
• Build and own the evaluation harness that tells us whether a change is actually an improvement. This is the backbone of the role.
• Diagnose where the agent fails at the frontier of its current capability and design the fixes.
• Set technical direction for agentic best practices internally; review, mentor, and raise the quality bar for the existing engineers.
• Translate latest research into pragmatic engineering decisions under real reliability and cost constraints.