Member of Technical Staff, Agent Platform

True Talent Solutions Partners•New York City, NY

14h•Onsite

About The Position

Join the applied AI team at a fintech company building agent-powered tools for financial advisors. You will own the agent platform, creating reliable systems that help advisors and clients handle complex tasks and collaborate effectively. The role involves building general-purpose and ambient agents, designing the agent harness that coordinates language models and related systems, and creating generative interfaces and human-in-the-loop workflows. You will also build evaluation infrastructure and memory systems that retrieve relevant client context and adapt to advisor preferences over time.

Requirements

Four to ten years of full-stack or backend software engineering experience, including building LLM-powered agent systems.
Production experience building AI agents with orchestration, tool use, and retrieval, and a track record of maintaining high-quality outputs.
Experience with LLM frameworks, RAG, prompt engineering, APIs, data modeling, and system design.
Familiarity with Python, TypeScript, React, vector databases, and agent frameworks.
Strong software engineering fundamentals and a track record of growth in a high-growth technology environment.

Nice To Haves

Experience shipping evaluation infrastructure for non-deterministic AI systems.
Experience with generative UI or human-in-the-loop interfaces.
A computer science degree.

Responsibilities

Build a general-purpose agent that plans and completes multi-step tasks using code execution, web search, and retrieval.
Develop ambient agents to triage email, process meetings, draft communications, and surface timely follow-up.
Design the agent harness that coordinates language models, context, tools, retrieval, and business logic.
Create generative interfaces and human-in-the-loop workflows that support collaboration between advisors and agents.
Build evaluation infrastructure for factual financial tasks and judgment-intensive work.
Develop memory systems that retrieve relevant client context and adapt to advisor preferences over time.