Senior Software Engineer in Test / SDET (AI-native)
Entwickelt und betreibt den KI-QA-Agenten, verantwortet Eval-Suites und Freigabe-Gates für LLM-Features und definiert Qualitätsstandards für KI-gestützte Entwicklung.
Entwickelt und betreibt den KI-QA-Agenten, verantwortet Eval-Suites und Freigabe-Gates für LLM-Features und definiert Qualitätsstandards für KI-gestützte Entwicklung.
Strategic Technical Partnership: Partner with product and design as a technical peer - shaping scope, challenging assumptions, and influenci
We are looking for a Senior AI Engineer who treats LLMs as production software, not as magic.
Company team.blue is an ecosystem of 60+ successful brands working together across 22 European countries to provide its 3.3 million SMB cust
Sprachmodelle in Fachanwendungen integrieren: RAG mit Retrieval und Reranking, Tool-Calling, MCP-Server, Multi-Agent-Abläufe Evaluierung auf
Wir schreiben dir, sobald eine neue geprüfte Stelle zu Large Language Models in Berlin dazukommt. Höchstens eine Nachricht am Tag, und nur, wenn es tatsächlich etwas Neues gibt.
Abmelden mit einem Klick, jederzeit. Keine Weitergabe an Dritte. Datenschutz
Du arbeitest in Kundenprojekten bei großen Unternehmen und öffentlichen Auftraggebern, eingebettet in deren Teams, mit voller Verantwortung für deinen Bereich.
Du arbeitest in Kundenprojekten bei großen Unternehmen und öffentlichen Auftraggebern, eingebettet in deren Teams.
Du arbeitest in Kundenprojekten bei großen Unternehmen und öffentlichen Auftraggebern, eingebettet in deren Teams, mit voller Verantwortung für deinen Bereich.
Baut agentische KI-Prototypen end-to-end, validiert sie mit Qualitäts-, Latenz- und Kostenmetriken und überführt sie in Referenzimplementierungen.
Verantwortet einen Produktbereich von der Problemfindung bis zum Launch und prüft, wo LLMs und Agenten echten Nutzen stiften.
Why SDK Engineering at Langfuse Your work will have an outsized impact.
Migriert Kunden-Frontends zu Next.js, auditiert Performance und baut vor Ort produktive KI-Agenten mit TypeScript und AI SDK.
Design and ship features in backend services (Java and Spring Boot) and/or web (TypeScript), with AI services in Python or TypeScript where
This role will design and develop technology in partnership with other software engineers in the US, UK, Germany, and Japan.
be part of our Data Science team to further optimize our innovative product based on machine learning and artificial intelligence - work on
is linked with our office in Berlin, but we are flexible about hybrid or remote work in Germany.
At Cresta, the AI Agent team is on a mission to create state-of-the-art AI Agents that solve practical problems for our customers.
We’re on the lookout for an ambitious AI-Engineer to join our Engineering team at our office in Berlin.
Your mission Join our ranking team and build the intelligent systems at the core of Cosuno s platform.
Entwickelt und betreibt produktive LLM-Pipelines für Logistikdokumente, inkl. Prompt-Design, Evals, Monitoring und Modellauswahl.
Entwickelt theoretische Grundlagen und Sicherheitsgarantien für neue KI-Architekturen (Scientist AI) und beweist diese formal.
This role is for engineers who care about velocity and solving real customer problems.
on this role - Strong professional experience in Python software development - Proven experience building AI-driven or agentic systems, pref
Train, test, and ship models that power Peec AI’s recommendations - helping customers boost their visibility in AI search - Develop algorith
This role is foundational data science: building the foundations of tabular foundation models so a single model can solve data-science problems across the board.
Own the quality of our LLM-powered features: eval coverage per feature, validated datasets, a gate every new AI feature passes before it merges, and clear routing to the owning squad when an eval fails Coach squads to write and maintain evals and tests for their own features, so quality is built into every team instead of checked at the end Develop our AI QA agent and other autonomous systems as a product: benchmark specs it has to pass, regression protection for every change to the agent, cloud runs fast and affordable enough that engineers use it on every relevant pull request, and repeatable release checks moved to it so release testing focuses on judgment, edge cases and readiness Define, together with product, design and engineering, the risk tiers that decide how much verification a change needs, from automated checks to extended testing and human review Make CI fast, cost-efficient and trustworthy: runner and model cost, pipeline duration, and flaky tests fixed systematically Refine our quality standards in the skills and rules our coding and review agents follow, so recurring issues are prevented at the source Review specs for testability before development starts, including capabilities that ship through our MCP server or as agent actions Measure quality with metrics that hold up as delivery speeds up: escaped defects, their severity, and the reliability of our agents and evals alongside more traditional KPIs What is the Tech Stack? Core: TypeScript, Node.js, Vue.js, MongoDB Testing: Playwright and Playwright MCP/CLI (E2E and as the browser for our QA agent), Vitest, Mocha/Chai AI in our SDLC: Cursor (including cloud agents and automations) and Claude Code, agent skills and rules shared across both, layered LLM code review in CI AI evals: eval suites per LLM feature, supported by LangSmith, running on relevant changes and as scheduled full runs Infrastructure and observability: GitHub Actions, Docker, AWS, Datadog (including CI visibility), Mixpanel Collaboration: Notion, Slack, GitHub You’ll be a good fit if you. See yourself as a software engineer who specializes in quality: you write production-quality TypeScript, build tools other engineers rely on, and debug down to the root cause, from a flaky test to a production issue Build systems that compound rather than running the same manual loop again: the second time you fix something, you automate it, encode it in an agent skill or remove its cause Have a strong eye for UI and UX: you notice when a flow is confusing or inconsistent, and raise it while it is still a spec or a prototype Have built or owned evals for LLM features and can tell a bad model output from a bad dataset Have run AI agents for testing or review in a team, and know where they fail and how to close the loop to prevent drift Hold a high bar for anything that touches customer data, and can show how a system meets that bar Think in risk, evidence and cost rather than coverage percentages, and act on it proactively: you spot where quality, speed or spend is slipping, from model choice to CI minutes and agent runtime, and fix it Have changed how an engineering team works, not only what it tests, and want to keep reinventing how quality is done
Learning and development: Grow your skills with a €2,000 yearly development budget. Work-from-home stipend: A one-time €1,000 stipend to set up your home workspace. Competitive pay and equity: Competitive salary plus a pre-IPO equity package, so you share in the company s growth. Corporate
: Exclusive discounts and deals from our partner network. Vacation: 30 vacation days a year plus public holidays, with the option to upgrade to 40 days at adjusted pay. After 5 years at Leapsome, take a month-long paid sabbatical. Wellness
: Stay active with a fully sponsored Urban Sports Club M membership, and support your mental health with an OpenUp subscription that includes 1:1 therapy sessions, plus company workshops on mindfulness and digital wellness. Hybrid and flexible work: Spend 2 days a week with the team in our Berlin office, Tuesdays and Thursdays. Impact: Join a fast-growing HR platform trusted by more than 2,000 companies worldwide: you ll help shape how companies run HR and grow their people in one connected system. Culture: A diverse, international team that values open feedback and celebrates wins, big and small. Sponsored lunch: Lunch is on us every Tuesday and Thursday. Work from anywhere: Work from a different location for up to 2 months a year. Monthly team activities: Join Feierabend, a monthly budget to get-together in Berlin — hosted by employees, for employees. This role is based in Berlin and the salary range for this position is €70,000 - €105,000 + Bonus + Equity. View our full compensation framework here. Leapsome aims to pay generously and fairly. If the listed range isn’t working for your situation, please tell us early in the process, we’d rather have that conversation than lose a great candidate over a number we may be able to address. If your role or skillset isn’t covered in our compensation framework, it’s likely because we haven’t hired enough in that area yet. We’ll build a data-backed range before we make you an offer, and we’ll share our thinking with you as early as we can. This note is published for transparency. It reflects our compensation philosophy as of June 2026 and describes our general global approach — it is not a contract, offer, or entitlement to any salary, increase, or bonus, and may change at any time. Individual compensation is always governed solely by your written employment agreement and applicable local law.
Your Opportunity Leapsome builds most of its code with AI agents. Cross-functional squads of product engineers own their features end to end and ship at a pace that would have needed a far larger team a few years ago. Speed only counts if quality keeps up.
Our customers trust us with performance reviews, compensation and personal employee data, so every change has to earn its way to production. QA at Leapsome is a small, senior team inside Product & Engineering, led by our Engineering Manager for QA. We build and run the quality system the whole department works with: an AI QA agent engineers run against their pull requests and staging, eval suites that gate our LLM features, layered automated code review that assesses every change by risk, and the CI pipeline everyone depends on.
Engineers across squads contribute to these tools. We own their direction and the standards they enforce. We are working towards an AI-native SDLC, where every change gets the depth of verification its risk calls for.
Automated tests and AI review run on every change, with agent-driven testing and evals wherever they apply. Changes touching permissions, compensation data and other sensitive areas get dedicated human review and testing on top. You will help extend and mature that system, set the bar it has to meet, and define the evidence we need before we rely even more on it.
Interesse an dieser Position?
Die vollständige Anzeige und das Bewerbungsformular liegen beim Arbeitgeber.
Zur Originalanzeige91 geprüfte Stellen bei 66 Arbeitgebern, 18 davon ortsunabhängig. 45 in Integration & Automatisierung, 24 in der KI-Entwicklung.