Arca - Member of Technical Staff, Agent Platform

deCircle; New York (NY); On-site; USD 200,000 - 300,000; Full time

Job summary

arcawealth.ai is hiring a Member of Technical Staff to own the AI stack end-to-end on-site at Flatiron, NYC. The team focuses on building agents and interfaces that assist financial advisors and clients.

You will design and implement a general agent, ambient agents, and the harness that coordinates LLMs, tools, and data, with a strong emphasis on evaluation and reliability.

Qualifications

• 4–10 years of experience in full-stack or backend engineering.
• Experience building LLM-powered agent systems.
• Strong evaluation/instrumentation mindset for complex systems.

Responsibilities

• Own Arca's AI stack end-to-end.
• Build a general agent capable of multi-step tasks.
• Develop ambient agents for client/advisor interactions.
• Create the agent harness that orchestrates LLMs, context, tools, retrieval, and business logic.
• Develop Generative UI and human-in-the-loop interfaces.
• Build evaluation infrastructure for high-stakes financial data and subjective tasks.

Skills

Full-stack engineering

Backend engineering

LLM-powered agent systems

Evaluation infrastructure

Probabilistic systems

In-person collaboration

Job description

Source: Paraform - https://www.paraform.com/share/-arca/cmrwefqnt00000al0hakfqkrg · Company: arcawealth.ai

Location: On-site, 5 days a week in Flatiron, New York City · Full-time · Salary $200K - $300K plus competitive equity · Open to visa transfers (e.g. OPT, H1B transfers)

The role

As a Member of Technical Staff focused on Applied AI, you'll own Arca's AI stack end to end. One framing the team keeps coming back to: agents are the primary users of the system of record. Everything built (the data models, the APIs, the UI) has to work for a non-human user that operates at scale, across every client, all the time.

What you'll build (and own)

• A general agent capable of complex, multi-step tasks (planning, sandboxed code execution, web search, retrieval) that powers a "do anything" experience for advisors.

• Ambient agents that act on behalf of clients and advisors: triaging email, processing meetings, drafting communications, surfacing what needs attention before anyone asks.

• The agent harness that orchestrates LLMs, context, tools, retrieval, and business logic into something coherent and reliable.

• Generative UI and human-in-the-loop interfaces where the agent and advisor genuinely collaborate, not just take turns.

• Evaluation infrastructure that holds two bars at once: high-correctness financial data and subjective, judgment-heavy tasks.

Example problems

• Human / AI collaboration in practice: how the agent updates its model of what an advisor wants when they override a recommendation, and learns what to do autonomously versus flag.

• Memory systems that know a client the way a great advisor does, across years, surfacing the right context at the right moment.

• Generative UI as an agent architecture problem: managing state as the agent updates the advisor's screen in real time, without the experience feeling unpredictable.

• Evaluations for financial services: high-accuracy financial data where errors have real consequences, plus judgment-heavy tasks, under compliance and auditability requirements.

Team

A 12-person team, one office, five days a week, working on problems that don't have clean answers.

• 4 - 10 years of experience in full-stack or backend engineering, building LLM-powered agent systems

• Comfortable where "correct" isn't always obvious; energized by probabilistic systems and rigorous about evaluation infrastructure

• Careful with consequences: agents here touch real retirement savings and estate plans

• Moves fast and treats evaluation and iteration as part of shipping

• High bar for quality; spots AI slop instantly and leaves systems better instrumented than found

• Kind, direct, low-ego; gives candid feedback and thrives in an in-person team