Lead AI Systems Engineer — Meridyn (AI implementation agency)

**Twyce Holdings is hiring for Meridyn, our AI implementation agency.** Meridyn builds production AI systems for businesses: agent workflows, LLM pipelines, and the integrations and operating controls that make them reliable in the real world. We are hiring the technical owner of delivery — not a narrow automation specialist. You will join client audit calls with the founder, turn business bottlenecks into sound technical plans, own architecture and delivery, and stay accountable for quality once systems are live. As the agency grows, you will set the engineering standard and help select and lead additional builders. **What you will own** - Production agent and LLM systems: tool use, multi-step orchestration, MCP, RAG, structured outputs, guardrails, and human-approval gates. - Integrations across APIs, CRMs, calendars, messaging, webhooks, databases, and undocumented client systems. - The operations layer: evals, tracing, prompt and version control, cost and latency management, safe releases, monitoring, and incident response. - A reusable delivery library in company-owned repositories, with documentation and QA built into every engagement. - Technical scoping on paid audit calls, and clear communication with non-technical clients. **What strong looks like** - 3+ years shipping production software, and at least 1 year operating LLM/agent systems beyond demos. - Strong Python or TypeScript; API and data-integration fluency; Git, SQL, deployment, testing, and secure secrets and data handling. - Evidence of a production AI system you have run for 3+ months — including what failed, what you measured, and what it cost. - Fluent, client-ready English, sound judgment, and documentation discipline. **First 90 days** 1. Take technical control of the live delivery backlog and establish a verified asset, access and system register. 2. Stabilise and document current client systems, with clear owners, monitoring, release checks and incident paths. 3. Deliver one complete client lifecycle, then turn the proven parts into reusable build packs and operating standards. **How we work** Remote and async-first, with meaningful UK/Europe overlap. An internal agent fleet removes admin and accelerates research, implementation and QA — but you own engineering judgment and every production outcome. Code, infrastructure accounts and documentation stay business-owned, so no individual becomes a single point of failure. **Process** Intro call, then a paid trial task (~4 hours, paid at £150–250), then a 30-minute live session where you extend your own submission, then terms.