Senior AI Agent Engineer — LLM Agents & Systems

Sharp Decisions; Jersey City (NJ); Hybrid; USD 180,000 - 210,000; Full time

Job summary

Sharp Decisions is recruiting for a senior, hands-on AI engineer to join the technology organization of a major global bank. This is a forward-looking role focused on agentic AI: designing, coding, and deploying autonomous agents built on large language models.

What sets this role apart is the problem space. The team is looking for someone who understands not just how to build an agent, but how agents behave once they interact with each other, with data, and with the institution around them.

Qualifications

• Deep understanding of AI agents: how they work, how they reason and act, and how they fail.
• Strong working knowledge of LLMs and building agents on top of them.
• Solid coding ability in Python, with PySpark and SQL for working with data at scale.
• Experience with Microsoft Azure.
• PhD in computational social science or similar interdisciplinary programs is a meaningful plus.

Responsibilities

• Design and build LLM-powered agents and multi-agent systems, from prototype through production.
• Write the code yourself: agent logic, tool use, orchestration, and the data pipelines that feed them.
• Model how agents interact within larger systems and identify where emergent behavior can create risk.
• Bring a complex-systems lens to how interconnected processes, models, and data flows affect one another.
• Work with engineering and business stakeholders to shape how agentic AI is applied responsibly across the bank.

Skills

AI agents

LLMs

Python

PySpark

SQL

Azure

Tools

Databricks

Job description

Sharp Decisions is recruiting for a senior, hands-on AI engineer to join the technology organization of a major global bank. This is a forward-looking role focused on agentic AI: designing, coding, and deploying autonomous agents built on large language models.

What sets this role apart is the problem space. The team is looking for someone who understands not just how to build an agent, but how agents behave once they interact with each other, with data, and with the institution around them.