Applied AI/ML Engineer (Agents)

CuspAIAmsterdamHybridFull-time14 Aug 2026

CuspAI is seeking an experienced Applied AI/ML Engineer (Agents) to design and build intelligent agents for our autonomous materials discovery engine. The role involves developing agentic frameworks, experimental design, and interdisciplinary collaboration with chemists and materials scientists. Ideal candidates have strong ML and software engineering skills, with experience in PyTorch or JAX, and a passion for solving global sustainability challenges.

What you'll do

Design the agentic framework that powers our platform to discover new materials
Build the integration that connects agents to ML models, simulation engines, databases, and heterogeneous compute backends
Design pipelines that let agents autonomously plan, schedule, execute, and interpret computational tasks at scale
Use expert annotations from the Chemistry team to drive targeted improvements in agent planning, retrieval, and decision-making
Create evaluations to measure the effectiveness of agents
Build agents that perform experimental design using Bayesian optimization, active learning, or related methods
Help close the loop between simulation and physical experiments
Develop strategies for multi-fidelity and multi-objective decision-making
Work closely with Chemists, Materials Scientists, and the rest of the Agent team
Work on customer projects and implement the direct needs required for these projects

What you need

Must-have

PyTorchJAXMachine LearningSoftware EngineeringCI/CDMLOpsLLM-assisted programmingAgentic frameworks

Nice-to-have

Materials scienceChemistryDrug discoveryAgentic frameworksLLM-powered applicationsReinforcement learningPlanning modelsSelf-improving systems

Soft skills

Proactive builder mentalityBias toward shipping and iterationWillingness to learnInterdisciplinary collaborationCommunication

AIMachine LearningAgentsMaterials ScienceRemoteFull-timeEngineeringTechnology

/What you'll do

/What you need

Must-have

Nice-to-have

Soft skills

What you'll do

What you need