Must haves

• 4+ years of professional software engineering experience with a track record of shipping production systems end to end.
• Real comfort working directly with customers, from scoping an ambiguous problem with a category manager to running a training session to answering a hard question from a chief procurement officer or CIO. You treat time with stakeholders as part of the engineering work rather than a detour from it.
• Hands-on experience building with LLMs in production: agentic workflows, tool use, retrieval, structured outputs, and evaluation.
• Strong backend and integration skills. API design, data pipelines, and working against systems whose documentation is wrong. Depth matters more to us than experience in any particular language.
• Experience shipping against messy real-world data: cleaning, normalization, entity resolution, and establishing ground truth.
• Judgment about what stays bespoke and what becomes product. Custom work is how we learn, but a deployment that only works for one customer is a liability.
• Comfort with early-stage ambiguity: a small team, shifting priorities, and shipping before everything is fully specified.
• Willingness to travel to customer sites, primarily manufacturing and industrial companies, roughly once or twice a month depending on the deployment.

Nice to haves

• Experience integrating with enterprise procurement or ERP systems such as SAP, Ariba, Coupa, Oracle, or NetSuite.
• Background in procurement, supply chain, sourcing, or manufacturing software.
• Deep production experience with TypeScript.
• Experience designing agent harnesses: the scaffolding that makes a model reliable in production.
• Experience with ClickHouse, or with other columnar analytics stores such as BigQuery, Snowflake, or Databricks.
• Experience meeting SOC 2, GDPR, or similar compliance requirements at a startup.
• Prior forward deployed, deployment strategist, solutions engineering, or technical consulting experience at a company like Palantir, Scale, or a comparable enterprise AI team.
• Experience at an early-stage startup (seed through Series B).

How you will make an outsized impact

• Customer obsession over elegance. You spend as much time understanding why a category manager does something the hard way as you do building the fix. The best architecture that nobody uses is worth nothing.
• Ship, then learn. You get something in front of a real user quickly and let their reaction shape the next version. Progress is measured in what got deployed, not what got planned.
• Own the outcome. Your responsibility doesn't end at the code you shipped. If a deployment is stalled on a data access request or a procurement team that hasn't been trained yet, unblocking it is part of the job.
• Generalize deliberately. You know when to hardcode something to unblock a customer this week, and you know when to stop and build the abstraction. You leave the platform better than you found it.