Senior AI Software Engineer

Über diese Stelle

<div class="content-intro"><p>At NiCE, we don’t limit our challenges. We challenge our limits. Always. We’re ambitious. We’re game changers. And we play to win. We set the highest standards and execute beyond them. And if you’re like us, we can offer you the ultimate career opportunity that will light a fire within you.</p></div><p><strong>The team</strong></p> <p>NiCE Labs (nice.com/nice-labs) is where NiCE explores what's next in AI for customer experience. This role is on AI Innovation, the Labs pillar that builds and validates new AI capabilities through prototypes, so the company can make better product decisions, faster.<span data-ccp-props="{"335559739":160}"> </span></p> <p><span data-contrast="auto">We're a deliberately small, senior team of builders inside NiCE Cognigy, the AI agent platform. We operate in fast iteration cycles: spot an emerging capability, build a working prototype, validate it against real metrics (quality, latency, cost, user impact), and then decide: kill it, iterate, publish it, or hand it off to a product team. Some of our work is public, like the open-sourced Cognigy Platform MCP server; most of it shapes the product roadmap from the inside.</span><span data-ccp-props="{"335559739":160}"> </span></p> <p><span data-contrast="auto">We hire for product mindset and builder mentality. Everyone on the team takes ideas from zero to working software, shows them to real users, and argues from evidence.</span><span data-ccp-props="{"335559739":160}"> </span></p> <p><strong>So, what’s the role all about?</strong></p> <p>You'll work at the intersection of frontier AI capabilities and a real enterprise product. The field moves faster than any roadmap can track. Your job is to close that gap: turn new models, protocols, and agent patterns into working systems, prove what matters in realistic scenarios, and produce the insights and reference implementations that let product and engineering teams move with confidence.<span data-ccp-props="{"335559739":160}"> </span></p> <p><span data-contrast="auto">You'll need to be comfortable with ambiguity, at home where best practices don't exist yet, and quick to move on when the results point elsewhere.</span><span data-ccp-props="{"335559739":160}"> </span></p> <p><strong>How will you make an impact?</strong><strong> </strong><strong> </strong></p> <ul> <li>Build agentic systems and full-stack prototypes end-to-end, shipping a first version in days and learning from real use</li> <li><span data-contrast="auto"> Track the frontier (new models, agent patterns, protocols like MCP) and turn the promising ones into working prototypes rather than slideware</span></li> <li><span data-contrast="auto"> Design experiments that show whether a concept holds up, measured in quality, latency, cost, and user impact</span></li> <li><span data-contrast="auto"> Work directly with internal users, product managers, and customers to validate concepts early and iterate quickly</span><span data-ccp-props="{"335559739":100}"> </span></li> <li><span data-contrast="auto"> Turn validated prototypes into reference implementations and write-ups that product teams can build on, then hand off cleanly and move to the next bet</span></li> <li><span data-contrast="auto"> Give product and platform teams concrete feedback on where AI capabilities shine and where they fall short</span></li> <li><span data-contrast="auto"> Move between initiatives as priorities shift. What you learn on one bet compounds into the next</span></li> </ul> <p><strong>Have you got what it takes?</strong><strong> </strong></p> <ul> <li>Have 4+ years building full-stack software, including things you started from nothing. We weigh what you've shipped more than years on a CV<span data-ccp-props="{"335559739":100}"> </span></li> <li><span data-contrast="auto"> Have shipped LLM-powered systems to production:features real users depended on, not demos or notebooks</span><span data-ccp-props="{"335559739":100}"> </span></li> <li><span data-contrast="auto"> Understand AI agents below the framework level: message arrays, tool calling, context management, prompt caching. You could build an agent loop from scratch and explain why it works</span></li> <li><span data-contrast="auto"> Use AI-assisted development (Claude Code, Codex, Cursor, or similar) as your default way of working, and it makes you measurably faster, not just busier</span></li> <li><span data-contrast="auto"> Are full-stackin practice: comfortable taking a prototype from API to demo UI on your own (our stack centers on TypeScript/Node.js and React)</span><span…