{"id":79214,"title":"Staff Software Engineer — AI Platform (all genders)","company":"TradeLink","city":"München","region":"Bayern","location":"Munich","work_model":"hybrid","employment_type":"full_time","employment":["full_time"],"roles":["data_science"],"visa":"not_mentioned","relocation":"not_mentioned","posted_at":"2026-10-23T12:36:22.263000+00:00","tools":["LLM","Ruby","Rails","React","TypeScript","PostgreSQL"],"seniority":"lead","years_experience_min":7,"german_requirement":"required","german_level":"unspecified","english_requirement":"required","english_level":"unspecified","team_language":"unknown","confidence":0.9,"method":"rules","language_badge":"German required","language_evidence":"Fluent German and English language skills are required","requirements_snippet":"We are looking for an experienced software engineer who has gone deep on AI, not someone trying to become an engineer through AI. Roughly 7+ years of professional software engineering experience, or equivalent depth. You have …","url":"https://jobs.eu.lever.co/tradelink/87a942c5-a0b2-4354-be23-85d74460f758","source":"lever","agency":null,"description_text":"About TradeLink\nTradeLink builds agentic platforms that lift European supply chains to the next level. Intelligent AI agents take over the routine, while the important decisions and priorities stay with your team. A network connecting companies, suppliers and carriers ensures maximum transparency, reduces lost time, and creates a secure, trusted process. With TradeLink, you're always in command of your logistics — not chasing it.\n\n\nBuild AI that does work — not just talks about it\nA supplier changes a delivery date in an email. The new quantity sits in a PDF. The operational system still contains yesterday's information.\nTradeLink builds AI agents for real supply-chain operations: agents that understand incoming information, use tools, update workflows and bring humans in when judgment or approval is needed.\nThe hard part is not getting an LLM to produce a plausible answer. It is building the harness around it so an agent can safely do useful work in a real business.\nThat is what you will build.\n\nYour responsibilities\nAs our Staff Software Engineer — AI Platform (all genders), you will build the common runtime behind TradeLink's agentic products.\n\nYou will own:\n\n•\nAgent context, tools, state, memory, retries and resumability.\n\n•\nPermissions, tenancy, validation, human approvals and audit trails.\n\n•\nEvals and observability so we can reproduce, measure and debug agent behavior.\n\n•\nRobust patterns for email, documents, APIs, browser-based systems and TradeLink actions.\n\n•\nThe line between deterministic application logic and model-driven decisions.\n\n•\nTurning successful customer workflows into reusable platform capabilities.\n\nYou will work on questions such as: How does an agent recover when a target system fails halfway through a task? What happens when two messages contradict each other? Which actions can happen autonomously? How do we know an agent change is safe before it reaches production?\n\nThis is a hands-on Staff IC role. You will write production code.\n\nWhat We’re Looking For\nWe are looking for an experienced software engineer who has gone deep on AI, not someone trying to become an engineer through AI.\n\n•\nRoughly 7+ years of professional software engineering experience, or equivalent depth.\n\n•\nYou have owned production systems: architecture, APIs, data models, reliability, observability and incidents.\n\n•\nYou have operated at Senior/Staff/Principal level, as a strong Founding Engineer, or with equivalent technical ownership.\n\n•\nYou have built LLM or agent workflows that actually run in production — beyond prompts and API calls.\n\n•\nYou understand tool calling, structured outputs, state, retries, evals, permissions and human-in-the-loop patterns.\n\n•\nYou have strong backend fundamentals: databases, APIs, authentication/authorization, background jobs and failure handling.\n\n•\nCoding agents are already a meaningful part of how you work.\n\n•\nYou still enjoy writing substantial production code yourself.\n• Fluent German and English language skills are required\n\nOur core product is built with Ruby on Rails, React, TypeScript and PostgreSQL. Rails experience is useful, but great engineering judgment and learning speed matter more.\n\nBackgrounds that could fit\n\nYou might be:\n\n•\nA Staff/Principal Backend or Platform Engineer from a strong SaaS company who became obsessed with agents.\n\n•\nA Founding Engineer from an AI-native startup.\n\n•\nAn Applied AI Engineer who is, underneath the title, an excellent software engineer.\n\n•\nA platform engineer from workflow automation, fintech or another environment where software actions have real consequences.\n\nYou do not need to be an ML researcher, train foundation models or have a PhD.\n\nThis probably isn't for you if\n\n•\nYour AI experience stops at prompting.\n\n•\nYou mainly want to research models rather than ship product.\n\n•\nYou prefer architecture documents to production code.\n\n•\nYou need detailed tickets before making progress.\n\n•\nYou want to build a large platform before proving what customers need.\n\n•\nYou believe generated code is useful without understanding and verifying it.\n\nHow we think about AI\n\nAI should multiply an already strong engineering foundation. We expect you to delegate aggressively to agents — exploration, implementation, testing, debugging — while remaining responsible for the result. You know where agents are excellent. You also know where they confidently produce nonsense.\n\nVerification matters as much as generation.\n\nWhy TradeLink\n\nWe already sit inside real supply-chain operations. We are not searching for hypothetical AI use cases. Every day, important processes still depend on people reading emails, opening PDFs and Excel files, calling for confirmation and manually transferring information between systems.\n\nWe are building an agentic layer around that work.\n\nYou get real production problems, direct access to Product and customers, and enough ownership to materially shape how the platform is built.\n\nHow we interview\nWe want the interview to resemble the job. We will talk about systems you have actually built, and in the technical working session AI tools are allowed and expected. We care about how you frame the problem, use AI, make decisions, verify the result and simplify.\n\nOne question before you apply: What do you do fundamentally differently as an engineer today because coding agents exist?\n\nWhat we offer\n• Impact, Innovation & Ownership: Shape our fast-growing, AI-native B2B startup in logistics. We work in small, focused pods, where everyone owns real outcomes – not just tasks. Influence product, customer solutions, growth, and brand. We think big, move fast, and solve real problems.\n• AI-Powered, Human-Led: Work at the forefront of an organization built AI-native. You use AI to multiply your impact – while judgment, creativity, and ownership stay clearly human.\n• Learn & Grow Fast: Continuous feedback, 360° reviews, and a dedicated development budget to build exactly the skills you need. Work closely with an experienced management team that sets direction – rather than managing your day-to-day in detail.\n• Passionate Team: Join a motivated, supportive team that values clarity, ownership, and real impact. In our pods, you work in tight-knit, well-coordinated teams – with real collaboration, not just shared tools.\n• Culture & Belonging: Inclusive, diverse, and fun – from virtual team events to unforgettable offsites, we build a strong sense of community everywhere, including regular in-person co-working in Berlin and Munich.\n• Own a Piece of Success: Through our VSOP program, you directly share in the company's growth.\n\n💙 Furthermore, you can read more about us and our culture here.\n \nIf you aren't completely confident that you meet our criteria, we recommend just taking the chance and applying anyway! Curiosity, passion for digital disruption, eagerness to learn, a collaborative spirit, and a growth mindset are most important to us.\n \n‍TradeLink strongly believes in the power of diversity and is highly committed to providing a respectful, safe, and welcoming environment for everyone who works here or with us. In order to eliminate unconscious biases in the recruiting process, you can remove the photo from your application. \n\nFor more information on data processing in the context of applications, please see our privacy policy for the TradeLink application system.","responsibilities_text":"As our Staff Software Engineer — AI Platform (all genders), you will build the common runtime behind TradeLink's agentic products.\nYou will own:\n•\nAgent context, tools, state, memory, retries and resumability.\n•\nPermissions, tenancy, validation, human approvals and audit trails.\n•\nEvals and observability so we can reproduce, measure and debug agent behavior.\n•\nRobust patterns for email, documents, APIs, browser-based systems and TradeLink actions.\n•\nThe line between deterministic application logic and model-driven decisions.\n•\nTurning successful customer workflows into reusable platform capabilities.\nYou will work on questions such as: How does an agent recover when a target system fails halfway through a task? What happens when two messages contradict each other? Which actions can happen autonomously? How do we know an agent change is safe before it reaches production?\nThis is a hands-on Staff IC role. You will write production code.","requirements_text":"We are looking for an experienced software engineer who has gone deep on AI, not someone trying to become an engineer through AI.\n•\nRoughly 7+ years of professional software engineering experience, or equivalent depth.\n•\nYou have owned production systems: architecture, APIs, data models, reliability, observability and incidents.\n•\nYou have operated at Senior/Staff/Principal level, as a strong Founding Engineer, or with equivalent technical ownership.\n•\nYou have built LLM or agent workflows that actually run in production — beyond prompts and API calls.\n•\nYou understand tool calling, structured outputs, state, retries, evals, permissions and human-in-the-loop patterns.\n•\nYou have strong backend fundamentals: databases, APIs, authentication/authorization, background jobs and failure handling.\n•\nCoding agents are already a meaningful part of how you work.\n•\nYou still enjoy writing substantial production code yourself.\n• Fluent German and English language skills are required\nOur core product is built with Ruby on Rails, React, TypeScript and PostgreSQL. Rails experience is useful, but great engineering judgment and learning speed matter more.\nBackgrounds that could fit\nYou might be:\n•\nA Staff/Principal Backend or Platform Engineer from a strong SaaS company who became obsessed with agents.\n•\nA Founding Engineer from an AI-native startup.\n•\nAn Applied AI Engineer who is, underneath the title, an excellent software engineer.\n•\nA platform engineer from workflow automation, fintech or another environment where software actions have real consequences.\nYou do not need to be an ML researcher, train foundation models or have a PhD.\nThis probably isn't for you if\n•\nYour AI experience stops at prompting.\n•\nYou mainly want to research models rather than ship product.\n•\nYou prefer architecture documents to production code.\n•\nYou need detailed tickets before making progress.\n•\nYou want to build a large platform before proving what customers need.\n•\nYou believe generated code is useful without understanding and verifying it.\nHow we think about AI\nAI should multiply an already strong engineering foundation. We expect you to delegate aggressively to agents — exploration, implementation, testing, debugging — while remaining responsible for the result. You know where agents are excellent. You also know where they confidently produce nonsense.\nVerification matters as much as generation.\nWhy TradeLink\nWe already sit inside real supply-chain operations. We are not searching for hypothetical AI use cases. Every day, important processes still depend on people reading emails, opening PDFs and Excel files, calling for confirmation and manually transferring information between systems.\nWe are building an agentic layer around that work.\nYou get real production problems, direct access to Product and customers, and enough ownership to materially shape how the platform is built.\nHow we interview\nWe want the interview to resemble the job. We will talk about systems you have actually built, and in the technical working session AI tools are allowed and expected. We care about how you frame the problem, use AI, make decisions, verify the result and simplify.\nOne question before you apply: What do you do fundamentally differently as an engineer today because coding agents exist?","status":"open","duplicate_of":null,"first_seen":"2026-11-04T00:12:52.812806+00:00","responsibilities_summary":null,"requirements_summary":null,"all_tools":["LLM","Ruby","Rails","React","TypeScript","PostgreSQL"],"posting_language":"en","evidence":[{"language":"german","requirement":"required","quote":"Fluent German and English language skills are required","source":"rules"},{"language":"english","requirement":"required","quote":"Fluent German and English language skills are required","source":"rules"}],"years_experience_max":null,"mobility_evidence":[],"model_version":"rules-8","english_badge":"English required","company_id":1639}