Summary
A technology team is looking for a full-stack AI engineer with 3+ years of production AI experience to build LLM-powered agents using tools and RAG. You will own evaluation suites, prompt engineering, and model-selection decisions, while also building dashboards, integrations, authentication, billing, and the supporting infrastructure. Strong Python or TypeScript skills are required, along with hands-on experience across at least two major LLM platforms or local models. Public technical writing or open-source work is also expected.
Highlights
Build production-grade AI agents on real-world data, own evaluation systems and model-selection decisions, work across the full application stack, and have visible impact on client-facing AI products and infrastructure.
Description
Design, build, and ship LLM-powered agents on real production data.
You own evals, prompt engineering, and the model-selection calls that make pilots succeed or fail.
What you'd ship
Build production agents (LLM + tools + RAG) for client operationsWrite and own eval suites — your work outlives every model migrationDrive vendor selection: Anthropic vs OpenAI vs open-source, by use casePair with consulting on discovery; write the ADR before the code
Must-haves
3+ years shipping AI in production (not just notebooks)Comfort across at least two of: Anthropic, OpenAI, Vertex, local modelsStrong Python or TypeScript; you've written evals from scratchPublic writing or open source we can read
Build the application layer around the agent — dashboards, integrations, auth, billing.
You own the surface clients see and the infrastructure that keeps it up.
What you'd ship
Next.js / TypeScript / Postgres / Vercel — end to endCustom integrations with client tools (Salesforce, HubSpot, Zendesk, custom)Build internal tooling that makes the consulting + AI team fasterOwn observability, CI/CD, and the on-call rotation
Must-haves
4+ years shipping full-stack at production scaleStrong React + TypeScript; you reach for SSR / RSC by defaultComfortable owning infra (Vercel, Postgres, queues, jobs)Pragmatic — you'll cut a feature to ship the deadline
Own top-of-funnel: paid acquisition, lead-gen experiments, conversion, attribution.
You run experiments end-to-end, from hypothesis to pipeline impact, with a real budget.
What you'd ship
Build and optimize paid acquisition (LinkedIn, Google, programmatic)Design and run experiments on hero, pricing, contact, applyOwn attribution model + pipeline reporting from first touch to closed-wonTight loop with sales — every campaign tied to qualified meetings
Must-haves
3+ years running paid acquisition for B2B SaaS or servicesYou've owned a marketing budget ≥$30k/mo and can show the dashboardFluent in attribution (UTM, dbt, GA4, or your own stack)Strong on conversion copy — you can ship the headline yourself
Compound organic distribution through long-form, technical content and GEO/AI-search optimization.
You own the pillar guide library, industry pages, and ranking strategy.
What you'd ship
Write or commission deep technical content on AI implementationOptimize for both Google and AI-answer engines (ChatGPT, Claude, Perplexity)Build and maintain a programmatic SEO surface (industries, services, use-cases)Pair with founders on thought leadership — long-form, signed pieces
Must-haves
Strong writing portfolio with technical depth (engineering-adjacent)You've moved a domain from
