Senior AI Product Engineer

About the role

Role Overview We're looking for a Senior AI Product Engineer to build the AI operating layer behind our marketplace. You take a valuable but unclear business problem — improve lead quality, cut refunds, automate a manual workflow, extract intelligence from calls — and turn it into a dependable production system end-to-end: investigate, define success, design, build, deploy, and improve on real usage data. You are a one-person AI product team for high-leverage problems, not a spec-taker. We are expanding beyond real estate into new service-business niches, and growth is powered by reusable software and automation — not proportional growth in manual work. Every system you ship either compounds into that leverage or exposes where it leaks — no in-between.

Responsibilities

• Discover high-value product and operational problems by working directly with users, business owners, and internal teams, then frame clear success criteria before writing code.
• Design and ship customer-facing AI features and internal automations across lead quality, matching, prediction, call intelligence, CRM, support, and operations.
• Build agents and workflows that safely use tools, APIs, databases, calls, SMS, email, and other company systems — with human approval where risk requires it.
• Create the context layer around models: instructions, retrieval, memory, structured outputs, tool definitions, permissions, and feedback loops.
• Build reliable event-driven and scheduled automation using webhooks, queues, schedulers, retries, idempotency, audit logs, and graceful fallbacks.
• Create evaluation datasets and automated checks that measure usefulness, accuracy, failure modes, regressions, latency, and cost — before and after release.
• Investigate real production failures and improve the complete system, not just the prompt or the model response.
• Turn successful solutions into reusable components and patterns that support additional iSpeedToLead niches without rebuilding from zero.
• Own architecture, quality, security, and business outcomes — not just the pull request.
• First 30 days: Understand and ship — learn the product, data, and workflows, identify the best first problem with evidence, and release one useful, low‑risk improvement in production.
• Days 31‑60: Own the workflow — launch one meaningful AI product or automation with evaluation, monitoring, clear ownership, and a human fallback — running against real marketplace volume.
• Days 61‑90: Prove and repeat — demonstrate measurable business impact (fewer manual hours, faster verification, lower refunds, better matching, higher conversion, or shorter time from idea to production), proven reliability, and turn the solution into a reusable pattern for future workflows or niches.

Requirements

• Strong record of building and shipping production software end-to-end, ideally in a SaaS, marketplace, or workflow product — with real users and real revenue on the line.
• Strong Python and/or TypeScript, plus solid backend fundamentals: APIs, databases, auth, data models, testing, and deployment.
• Hands-on experience shipping LLM-powered functionality using model APIs, structured outputs, tool/function calling, retrieval, context engineering, and agentic workflows.
• Production automation experience: webhooks, events, queues, schedules, retries, idempotency, monitoring, and failure recovery — not just happy‑path demos.
• Practical evaluation skills: test sets, traces, quality metrics, regression checks, feedback loops, and root