You own and measure success by agents live in production, adoption by the user group each was built for, and delivery inside the fixed fee we quoted. You work closely with vCAIOs, other Forward Deployed Engineers, Product and GTM.
This role is remote (US).
In this role you will
Join vCAIO-led opportunity assessments and find the workflows where data is spread across systems nobody reads in the same week
Scope each build: the systems it reads, the actions it takes, its user group, its success metric and baseline, and say so when the bottleneck sits outside the client’s control
Build the agents, connectors (Microsoft Graph, SharePoint, Teams, Outlook, client APIs, MCP servers) and skills each engagement needs
Deploy into client tenants with least-privilege access, an entry in the governance register, and documentation that holds up to an auditor or a DDQ
Measure each agent against its baseline, train its users, and hand it into ongoing maintenance under the platform
Turn what works in one engagement into connectors, skills, evals and templates the next one can reuse
A typical engagement
A first agent goes live in month three of a client’s program, at a fixed fee, measured against a baseline taken in month two.
Month 1, platform connected, shadow AI and spend baseline: review the tenant inventory; note the systems and data you will need
Month 2, vCAIO-led agent opportunity assessment: join interviews, score workflows on value, risk and effort, capture the baseline, write the scope
Month 3, first agent build: build, deploy to a named user group, measure, hand over
Month 4+, further agents scoped per project: maintain the live agent; scope and build the next one
Example projects
These are the kinds of agents you would build in your first months.
Medical device manufacturer, customer site activation tracker: reads the commercial team’s Teams channels, account files in SharePoint, Outlook threads and the current tracker; follows each new customer site through contracting, training and first use; flags sites that have stalled or lack a named owner. The commercial team sees every stalled site in one place, every week.
Hedge fund, investor DDQ responder: reads past DDQs, policies and the compliance library; drafts answers to a new questionnaire with a source for every line. First draft in hours instead of two weeks of IR and compliance time.
Industrial manufacturer, supplier quality monitor: reads supplier certificates, incoming inspection results from the ERP and nonconformance reports; flags expiring certifications and repeat defects; drafts the supplier corrective action request. Quality catches supplier issues before they reach the line, and every request leaves an audit trail.
You might thrive in this role if you
Bring 4+ years of engineering experience, including client-facing work
Have built or shipped LLM-powered agents and know how model behavior shapes the product experience
Write and review production-grade code in TypeScript, Python, or a comparable stack
Know Microsoft 365 as a data source: Graph API, SharePoint, Teams, Outlook, and how app permissions and consent work in a tenant
Can run a scoping session with a business leader, say no to a weak use case, and write a scope a CFO will sign
Would rather put a working agent in front of 10 users in three weeks than a perfect one in front of none in three months
Take extreme ownership of problems and delivery, and handle client data with a security mindset
Bonus
Forward deployed, solutions or sales engineering experience at a SaaS, AI or consulting company
Built on MCP: servers, connectors, or skills for Claude, ChatGPT or Copilot
Logistics
Location: Remote (US)
Travel: Up to 5%, for on-site discovery and launch
Compensation: $150,000–$200,000 base
Interview process: recruiter screen; technical call; take-home or live build of a small agent against a mock Microsoft 365 tenant; client scoping role-play with a vCAIO; founder conversation