AI Automation & Agent Engineer
At a glance
Mid-level AI Engineering role at Sportsdigita. New York · full-time · no salary stated · we estimate $157,938–$275,000.
Sportsdigita has not published a salary for this role. The figure below is our estimate, not theirs.
aimlroles summary, based on the employer's posting.
What you'll do
- Sit with departments to map manual work, then rank automation opportunities by time saved, fewer errors and revenue impact
- Build and maintain agents and workflows that connect Salesforce, Google Workspace, Slack and other named systems
- Oversee deployed AI tools, including access, model choices, spending and decisions about what to keep or retire
- Teach teams through sessions, documentation and coaching so they can create prompts, skills and simple automations
- Track results and brief executives quarterly on time recovered, departmental use, workflow costs and retired tools
What you bring
- Three to six years in software, automation or solutions engineering, or technical operations, with shipped work people used
- Experience with Anthropic’s API, Claude Code, Claude Skills or MCP servers
- Practical LLM-building skills, including prompt design, tool use, structured outputs, retrieval and production-quality evaluation
- Ability to work with business SaaS systems, especially Salesforce, and to assess security, permissions and customer-data risks
Who this fits
This role suits a self-directed engineer who can investigate unfamiliar teams’ workflows, explain technical tradeoffs to different audiences and decide when AI is not the right solution. It spans multiple departments and combines hands-on building with internal coaching, tool stewardship and executive reporting.
From the employer
AI Automation & Agent Engineer
We are hiring someone to own AI automation end to end. This is not a research role and it isn't a side project bolted onto another job. You will sit across Sales, Customer Success, Creative, Engineering, Finance, and Marketing, find the work that shouldn't be done by hand anymore, and build the agents, skills, and integrations that take it off people's plates.
Responsibilites
Embed with departments and find the work worth automating. Run discovery with each team, map their workflows, and build a prioritized backlog of automation opportunities ranked by hours saved, error reduction, and revenue impact. Push back when AI isn't the right answer — a better Salesforce validation rule sometimes beats an agent.
Build and ship internal agents and automations. Design, build, deploy, and maintain agents and workflows that connect the systems we already run on: Salesforce, Gmail and Google Workspace, Slack, Zoom, HubSpot, Notion, Asana, Maxio, and the DIGIDECK platform API. This includes writing Claude Skills, standing up MCP integrations, and building the glue code that makes those systems talk to each other.
Own the internal AI tooling stack. Be the person who knows what's deployed, what it costs, what's working, and what's stale. Manage access, model selection, spend, and the lifecycle of everything you build.
Enable the teams, don't just serve them. Run enablement sessions, write internal documentation, and coach people into building their own prompts, skills, and light automations. Success means AI fluency spreads past you, not that every request routes through you.
Measure and report on impact. Instrument what you ship. Bring a quarterly readout to the executive team on hours recovered, adoption by department, cost per workflow, and what you're retiring.
Set the guardrails. Establish practical standards for how we use AI with customer and company data — what goes where, what gets human review, how we handle credentials, what we log.
Requirements
- 3–6 years in software engineering, automation engineering, solutions engineering, or technical operations — with a track record of shipping things people actually used
- Experience with the Anthropic API, Claude Code, Claude Skills, or MCP servers
- Hands-on experience building with LLMs: prompt design, tool use and function calling, structured outputs, retrieval, and evaluating whether output quality is good enough to trust in production
- Fluency administering or integrating with business SaaS systems — Salesforce especially, plus the usual GTM and ops stack
- The consulting instinct: you can walk into a department you know nothing about, ask good questions, and leave with a clear picture of where the time goes
- Ability to explain technical tradeoffs to a CSM, a designer, and a CFO in three different registers
- Sound judgment about security, permissions, and customer data, you know what should never be pasted into a chat window
- Self-directed. You will be handed problems, not tickets.