Forward Deployed Architect

About the Role

Our agents build audiences, run journeys and activate campaigns in the time it used to take a marketer to file a ticket. Those agents rely on a solid data foundation and set of context, which is what this role is for. You go into our most complex enterprise customers and set the vision for their architecture, deploy a team of agents to build it, and you catch and solve any quality issues along the way.

You'll work with a small number of enterprise customers to figure out what they need, champion a specific implementation plan, and build it yourself or in conjunction with the customer's team. You'll define what success looks like, and you'll be the one who proves the job is done. This is a new role on a team we're building now. You'll be one of the first people in it, which means you'll shape how it works — the playbooks, the standards, what we take on and what we hand back.

Our platform does the heavy lifting. You drive the vision for the data architecture and have the judgment to dive into the details when the details are what matter. The best candidates are fluent in marketing outcomes and customer engagement, with real depth in data.

What You'll Do

• Establish what the customer is trying to do. Run discovery with marketing and data teams and get from "we want better targeting" to specific activation use cases.
• Profile and understand the data, with agents doing the first pass and you deciding what it actually means, so you can architect a solution that closes data gaps at the customer and lands them with a working Hightouch configuration
• Create target-state schemas and identity matching rules that make a customer’s data activation-ready
• Deeply understand how the customer targets and at what grain — e.g., person, household, device — and design identity matching rules to support it
• Direct the build. Put agents to work profiling the warehouse, proposing models and generating transformations. Review everything they produce, rewrite what they get wrong, and own the models and their performance regardless of who or what typed them.
• Define the number that proves it worked, agree it with the customer before the build starts, compute it, and defend it to senior leaders
• Hand off the day-to-day but stay the architect. You remain the named person they and their account manager come back to when the data model changes, new sources land, or their identity graph needs rework. You produce reference architectures the next engagement can reuse
• Build the context our agents run on. The semantic layer, metric definitions and business logic you encode are what let a marketer at a Fortune 500 ask a question in plain language and get an answer worth acting on. Bad context is worse than no context, and it's yours
• Be the eyes and ears for our product teams. You'll see our gaps before anyone else does - bring them back sharp enough to act on

About you

• You direct agents, you don't just prompt them. You'll point them at warehouses with undocumented tables and no reliable dictionary, and you'll be the last judgment before what they build ships to millions of people. You reach for an agent before a blank editor, but you scrutinize every line before it goes out.
• SQL Fluency. Anyone can generate a thousand lines of SQL now. You're hired because you can read them and find the join that's wrong. Fluent in Snowflake, Databricks or BigQuery, with query performance and warehouse cost as things you think about unprompted.
• You're a driver. You've parachuted into messy engagements with dozens of stakeholders who have differing opinions and driven discovery, clarity, and alignment.
• 5+ years in a customer-facing technical role — analytics engineering, solutions architecture, forward-deployed engineering, data consulting, or in-house marketing data work at a large brand.
• This is an agent-native role. The work is data architecture. The leverage is AI. If your instinct is that doing it by hand would be faster, this will be the wrong seat.

Bonus Points

• dbt, SQLMesh or similar transformation tooling; git-native, PR-driven workflow.
• Prior work with identity resolution, CDPs or MDM