Principal Cloud Engineer AWS/Multiple
•CloudscalerHybrid· 20h
💻 Engineering⏰ Full-time🎯 Lead / Principal💰 £95k-130k🇬🇧 Aldgate, UK
20h ago
Devsavant·19 days ago
DevSavant is an operating partner for startups and growth-stage companies, helping them turn ambition into execution.
We support founders and leadership teams with product engineering and global staffing, from early prototypes and MVPs to scaling high-performing teams. Our vetted talent across LATAM and Asia embeds directly into client teams, operating as true extensions rather than external vendors.
With over 8 years working in venture-backed ecosystems, DevSavant is trusted to accelerate delivery, scale teams efficiently, and support companies as they reach their next milestone.
We're looking for a Principal Engineer / Architect in AI Agents to own the architecture and technical direction of our AI agent platform end to end. This role sits at the intersection of deep systems engineering and frontier AI — you'll design the agent runtime, own the cloud infrastructure that runs it, build the evaluation harness that measures quality, and set the technical bar for a small but growing team. You'll be the operational and architectural center of gravity for a platform built to run AI agents reliably in front of real clients.
We run AI agents in production, and this role is how we make them excellent. If you're energized by the challenge of building systems that are robust, scalable, and observable enough to hold up under real conditions — and you want full ownership over a platform reshaping how agents are built and deployed — this is the role for you.
You will report directly to the Founder. This is a remote/hybrid role, and we are looking for candidates who bring fluent, polished, client-facing professional English.
Own the system architecture end to end — agent runtime, infrastructure, and delivery — including all trade-offs and design decisions, with no architect above you to defer to.
Drive the engineering that makes our agents robust, scalable, and observable enough to run reliably in front of clients; treat quality as a product surface, not an afterthought.
Build and maintain the eval harness that determines whether an agent is actually good — designing offline and online evaluations, LLM-as-judge scoring, and closing the loop between measurement and improvement.
Own how live agents are deployed, monitored, and recovered — including catching silent failures and ensuring full observability across the agent stack.
Set technical direction: bring proposals, decide what moves next, and drive the engineering roadmap in close partnership with the Founder.
Review work, mentor engineers, and grow the team as the platform scales — shipping low-risk decisions independently and bringing weight-bearing ones to the table.
7+ years of experience in software engineering, systems architecture, or a closely related discipline — with meaningful, production-grade time spent building and operating AI agent systems.
Agent-harness mastery is the core of this role. You've built, deployed, and operated agents on real harnesses in production — not in demos. You have working knowledge of OpenClaw and Hermes at the architectural and internals level: enough to extend, debug, and push them past their defaults. You're pragmatic about reaching for whichever harness fits the problem.
Fluent across the full agentic stack — tool and prompt orchestration, multi-agent workflows, function calling and structured outputs, RAG and agent memory, MCP, and multi-provider LLM integration — and you know the failure modes unique to agents: loops, silent stalls, context drift, tool misuse.
You treat evals as first-class engineering. You can design and implement evaluation frameworks, reason rigorously about agent quality, and instrument agents with proper tracing and monitoring.
Strong cloud and infrastructure depth — you've owned full systems from a blank page and designed for scale, reliability, security, and cost across AWS, GCP, or Azure, with solid command of containers, orchestration, infrastructure-as-code, and multi-tenant architecture.
A real DevOps / SRE backbone: live ownership, incident response, root-cause analysis, and distributed-systems instincts. You build production-grade systems that are scalable, fail-loud, and cost-aware — not impressive demos.
Strong leadership and judgment — you set direction, decompose ambiguity, mentor strong engineers, and communicate clearly enough to be trusted in front of clients.
You track the agent and LLM space obsessively — a new harness or model lands and you're already trying it.
You have experience with Go for infrastructure or systems work.
You've worked with local inference tooling such as Ollama or vLLM.
High autonomy, low overhead. We ship production-grade work fast, keep decisions close to the people doing the work, and pull in oversight only where it genuinely matters. If you do your best work when you own the problem and the bar is high, you'll fit here.