Engineering Applied AI Engineer Full‐time Remote (US) View role →

Topsail; United States; Remote; USD 140,000 - 210,000; Full time

Benefits offered by this job

Remote-first

Meaningful equity

Above-market compensation

AI tooling budget

Job summary

Topsail is hiring an applied AI engineer to design, build, and maintain agent runtimes, tool integrations, evals, and observability that power client deployments. You will ship capabilities across deployments, not just demos, and help scale patterns from one client to the next in a week, not a quarter.

You will work with forward-deployed engineers and strategists, turning field learnings into reusable infrastructure while remaining non-client-facing by default.

Qualifications

• Must have shipped an agent or LLM system used by real users.
• Strong opinions on agent architecture: context management, retrieval, and tool design.
• Solid software engineering fundamentals: backend, APIs, reliability, and cost efficiency.

Responsibilities

• Design and build agent systems, including planning, tool use loops, memory, guardrails, and human handoffs.
• Own the harness: evals, traces, and judge-in-the-loop patterns to benchmark performance.
• Build the shared platform: MCP servers, connectors to ERPs/CRMs, Salesforce, Excel, Slack, etc.
• Keep the system model-agnostic to allow move between major providers without rebuilds.
• Pair with forward-deployed engineers to turn one-off fixes into reusable infrastructure.

Skills

Shipped agent/LLM system

Backend software engineer

Model-agnostic deployments

Observability & runtime

Tools

Salesforce API

Slack integrations

Job description

Build the agent systems and the platform underneath every deployment. Not client‑facing by default. Deep, reusable, and measured.

Full‑time Remote (US) Occasional travel Reports to the founders

Topsail deploys agents inside growing companies, and the forward‑deployed team can only move as fast as the systems underneath them. An applied AI engineer here builds those systems: the agent runtimes, tool and MCP integrations, evals, and observability that every client deployment runs on. You are the reason a pattern that worked at one client ships at the next one in a week instead of a quarter.

This is not a client‑facing role by default. You work with the forward‑deployed engineers and strategists, take the hardest technical problems off their plates, and turn what they learn in the field into reusable infrastructure.

What you will do

• Design and build agent systems: planning and tool‑use loops, retrieval over a client's ontology, memory, guardrails, and handoffs to humans.
• Own the harness. Evals, traces, and the judge‑in‑the‑loop patterns that tell us whether an agent is actually doing the job, against a baseline.
• Build the shared platform: MCP servers, connectors to the systems clients run on (ERPs, CRMs, Salesforce, Excel, Slack), and the company brain that gives every AI tool the same governed context.
• Stay model‑agnostic. Abstract the model layer so a client can move between Anthropic, OpenAI, and Google without a rebuild, and benchmark when a new release lands.
• Pair with forward‑deployed engineers on the hardest deployments, and turn one‑off fixes into things the whole team reuses.

What we look for

• You have shipped an agent or LLM system that real users depend on. Not a demo. We will ask about the failure modes you found in production and what you built to catch them.
• You have strong opinions about agent architecture because you have been burned: context management, retrieval and chunking, hybrid search, tool design, evals, and the cost and latency tradeoffs of a deployment stack.
• You are a strong software engineer first. Fluent in a modern backend language, comfortable with relational data and APIs, and disciplined about making something reliable, observable, and cheap to run.
• You live in Claude Code, Codex, Cursor, or similar every day and can show us how.
• You write clearly. Your design docs are the ones other engineers actually read.

Nice to have

• Experience with MCP, agent SDKs, or building developer platforms.
• Serverless or edge runtimes, and event‑driven systems.
• Time inside a real‑economy business, or a taste for how one runs.

What you get

• Systems that run inside real businesses, owned by you end to end.
• A new industry and a new data problem every few months, next to the people who built the company.
• Remote‑first, with time in New York, San Francisco, Boston, and Lehi, Utah when it matters.
• A serious AI budget: frontier model access, the tools you want, and a lot of tokens.

Compensation: Above‑market salary and bonus, plus meaningful equity.