SlipstreamJobs tracks this role from the company's public career site. Apply directly on the employer's site.
Salary: USD 215,000 - 235,000 / annual
The Electric Plant Co. is building plant and tree intelligence using IoT hardware that measures electrical signals in plants paired with environmental data, decoded by a foundation model called Bombadil. The company translates raw physiological data into actionable insights for growers.
You will build the application layer between Bombadil and end users—the client context layer, insight cards, learning loops, agents, and scheduled work. You'll start embedded in the field with the Deployment Lead on the first real customer, shipping tools growers actually use to run their orchards. From month six onward, you'll convert field learnings into reusable protocols and tools that accelerate every subsequent deployment.
This is a senior individual contributor role reporting to the Director of Engineering with direct visibility to the CEO. You'll collaborate closely with the engineer owning the deterministic data plane, the foundation model team, and the client relationship owner.
You will own: the client context layer and retrieval (built and measured, not assumed); the tool schema exposing platform data and insights with uncertainty quantification; the product's agent systems as versioned, testable, reviewable artifacts; one insight pipeline rendering to chat, email, PDF, and API; the context schema and card structure enabling multi-tenant isolation and client IP protection; actuation (ensuring policies execute reliably, safely re-runnable, fully trackable, with proper approvals); and eval suites, trace capture, drift monitoring, and per-client cost accounting.
You should have shipped production AI retrieval and context assembly with quantified usefulness metrics (not demos). You consistently push repeatable work into deterministic code and spend model calls only where judgment is required. You've shipped multi-tenant isolation in production with tenant-scoped data, per-client credential scoping, and proper deletion. You've built config or skill systems that non-engineers used and modified daily. You've operated agent systems in production (not prototypes) with real stories about silent drift, cost blowups, or operational failures. You've built durable, idempotent, observable scheduled or event-driven agent work. You've built eval suites, trace capture, or drift monitoring and can speak to ground truth sourcing and arrival frequency.
The ideal candidate has a full-stack or front-end web/application engineering background with a deliberate turn toward agentic systems, rather than a backend or data engineer entering this space for the first time. Total years matter less than this trajectory.
Strong plus: integration grit against client and legacy systems (enterprise APIs, CSV over SFTP, email deliverability).
This role is not for someone seeking a settled spec and quiet backlog—requirements arrive mid-season and are often wrong. It's not a research role (you consume and expose the model, don't train it) and not infrastructure (the data plane is owned elsewhere). If you're more excited by orchestration toolkits than by getting something working in a grower's hands, this isn't the fit.
Requirements:
- Production experience with AI retrieval and context assembly, with quantified system usefulness
- Demonstrated ability to push repeatable work into deterministic code
- Shipped multi-tenant isolation in production (tenant-scoped data, per-client credentials, deletion)
- Built config or skill systems used and modified by non-engineers
- Operated agent systems in production with operational incident experience (drift, cost, failures)
- Built durable, idempotent, observable scheduled or event-driven agent work
- Built eval suites, trace capture, or drift monitoring with ground truth sourcing experience
- Full-stack or front-end web/application engineering background with deliberate pivot to agentic systems preferred
