AI Engineer, Agent Builder (Remote).

Catalyst Wayfare; United States; Remote; USD 120,000 - 170,000; Full time

Job summary

Catalyst·Wayfare AI is an AI transformation firm that ships production AI systems for mid-market enterprises in regulated and technical domains. This role sits in a remote-first environment with occasional U.S. travel and overlap with US Eastern time.

You will engineer end-to-end capability agents, integrate with client tools, and ship working systems rather than demos. Collaborate with a tiger-team of senior engineers and own artifacts from development to production.

Qualifications

• 3–5 years of professional software engineering experience with Python.

Responsibilities

• Build individual capability agents end-to-end: prompts, tool integrations, retrieval, evals, monitoring.
• Build on production data foundations: RAG system with 11,000+ knowledge chunks and live telemetry.
• Implement integrations between orchestration layer and client engineering tools via APIs.
• Run iteration loops with senior engineers to ship, fail, and improve.
• Create evals and monitoring dashboards to verify agent effectiveness.
• Contribute to the orchestration engine (state, error handling, audit logging).
• Document what you build so the client team can extend it after handoff.

Skills

Python

LLM APIs

Agent systems

APIs

Retrieval

Cloud infra

Monitoring dashboards

Data integration

Tools

Cursor

OpenAI API

Anthropic

Neon

Sentry

Job description

Remote-first, US Eastern overlap
Occasional U.S. domestic travel (major metro)

About Catalyst·Wayfare

An AI transformation firm that ships working systems.

Catalyst·Wayfare is an AI transformation firm that builds production AI systems for mid-market enterprises in regulated and technical domains. We do not stop at PowerPoint AI strategy. We ship working systems alongside our clients' engineering teams.

The team's backgrounds span MIT, McKinsey, the White House, and senior technology and policy roles in government.

Why this role

Ship agents engineers use on Monday morning.

You would embed alongside our Lead AI Engineer at a leading firm in a critical-infrastructure sector, shipping individual agents that real engineers use on live production work. Not demos that die in staging. The feedback loop is tight: skeptical, brilliant domain experts use what you build, break it, and tell you exactly why.

This is a center-of-the-action engineering seat. GitHub history, not calendar invites. You write the code; we manage the room. Not a junior role - a builder role, with autonomy on individual agents you own end-to-end, real production stakes, and clear paths to grow with the firm. The work you do here becomes the template we replicate at the next client.

How we ship

We use what we sell.

Claude and OpenAI APIs in production. Open-source models (Llama, Mistral, Qwen) when the data or the math points there. Cursor and Claude Code in our IDEs, daily.

Vercel, Neon, Sentry as the deploy surface. Modern infra, no six-month wait for IT to approve a tool you already use at home.

Evals are first-class artifacts, not an afterthought. Agents we trust live behind audit trails. You will not be the engineer fighting a CISO to install Cursor.

What you will do

Individual agents, end-to-end.

• Build individual capability agents end-to-end: prompts, tool integrations, retrieval, evals, monitoring.
• Build on existing data foundations: a production RAG system with 11,000+ knowledge chunks, established retrieval infrastructure, and live interaction telemetry.
• Implement integrations between our orchestration layer and client engineering tools via APIs the client team builds.
• Run iteration loops with the tiger team of senior engineers. Ship, watch them break it, fix, ship again.
• Build evals and monitoring dashboards so we can tell whether each agent is actually working.
• Contribute to the orchestration engine (state, error recovery, audit logging) under the Lead Engineer's direction.
• Document what you build well enough that the client's team can extend it after we hand off.

What you will bring

The shape of the right person.

Must-haves

• Three to five years of professional software engineering experience, strong Python.
• AI-pilled and full-stack. You reach for agents instinctively and have opinions about which ones. But you got fluent with code first - Cursor multiplies you; it did not teach you.
• Real production experience with LLM APIs (Anthropic, OpenAI, or similar). Not just demos. You know what context windows actually cost, how to handle structured outputs reliably, and why your retrieval is bad even when it looks good.
• You have built and shipped at least one agent system end-to-end — orchestration, tools, retrieval, evals — solo or as part of a tiny team. Something a real user touched.
• Comfortable with messy integration work: APIs, parsing, schema mismatches, brittle upstream systems.
• You iterate fast based on user feedback without getting precious about your code.
• Clear written communication and willingness to overlap with (not mirror) US Eastern time for core collaboration hours.

Nice-to-haves

• Multi-agent systems in production with measurable adoption, or significant open-source contributions to agent frameworks.
• Experience with retrieval and RAG systems beyond toy implementations.
• Cloud infrastructure (AWS, Azure, GCP) for deploying production AI applications.
• Background in technical or regulated industries (financial services, healthcare, legal, industrial).
• Ability to be on-site with the client occasionally during build sprints. Travel is to a single U.S. major metro.

Reporting

To the Lead Engineer.

Reports to the Lead AI Engineer for day-to-day technical direction, with founder oversight on engagement strategy. Client stakeholders are managed through us, not through you.