Skip to content
Search jobs inside your AI agent. Claude, ChatGPT, Claude Code, Codex, Gemini, Cursor and Grok
Connect

Sea Sentinel AI

NewBe an early applicant

Founding Backend & AI Engineer

Posted Updated

Summary

Founding engineer at Sea Sentinel AI (Munich, hybrid) who co-owns PRISM, the LLM-driven pipeline analyzing 800k+ daily news articles to detect maritime shipping disruptions. Day-to-day: production engineering, prompt tuning with evals, and expanding the platform into verticals like air cargo and energy risk.

Sea Sentinel AI · Munich · Full-Time



What we do


Sea Sentinel AI is a risk intelligence platform with a focus on the maritime shipping industry. We monitor ports, vessels, and shipping routes worldwide, detecting disruptions before anyone else does.

When a vessel crashes, we detect it from a local newspaper hours or days before major media picks it up. When a typhoon is still days away from hitting a port, we warn customers before the chaos starts. This way, we enable our customers to be ahead of crises, rather than reacting to them at the last minute. We sell our intelligence primarily via API, which means our customers integrate it directly into their own systems and workflows.

Our rigorous focus has allowed us to build the leading product in this niche. It shows: we consistently win deals with global customers, large and small. We grow because customers pay for what we build.


The role


Our core data infrastructure, PRISM, is built and proven. Every day it analyzes more than 800,000 news articles alongside weather and other data to find the disruptions that matter to our customers.

  • Owning PRISM. This is your first priority. Keeping PRISM sharp takes two things: solid traditional engineering, and serious prompt tuning backed by evaluations and test harnesses. You'll own this together with the team.
  • Taking PRISM into new verticals. We're ambitious. Next up are other verticals such as air cargo and energy risk intelligence (LNG, oil, gas). You'll identify the master data each vertical needs, benchmark data providers, and make the case for what to integrate and how to integrate it into PRISM. You'll work with structured and unstructured data and form your own views. This is where the role becomes product work: you care whether what you build solves the right problem.
  • Working on a system that’s still moving. PRISM runs on messy, real-world data, and new models change what’s possible every few months. You are encouraged to test your ideas against ambiguous inputs and production conditions where things break, and you’ll decide which new model capabilities are worth adopting. Then you’ll prove it with evals.

You won't have a backlog handed to you on Monday morning. You'll be expected to understand the system well enough to know what the most important problems are and go solve them.


What we're looking for


Must haves:

  • You've run something in production and dealt with it breaking. You know what silent failure looks like and how to find it, using sensible monitoring and alerting.
  • Good coding standards and the judgment to know when to apply them and when to move fast.
  • You drive your own work: you identify problems, prioritize, and see them through without someone telling you what's important.
  • Analytical curiosity. You're comfortable sitting with ambiguous questions, forming hypotheses, and making a case for your conclusions.
  • Real experience tuning LLM systems. You know from experience that a prompt which works well on one model can fall apart on another, or on the same model with slightly different input or settings. You treat model behavior as something to measure, never something to assume.
  • An evaluation mindset. Fixing one case often breaks another, so you build eval sets and regression tests, measure before and after, and only ship a change you can show is a net improvement.
  • Experience monitoring and tracing LLM pipelines in production. The question you should be able to reason about: if the input changes, or we turn this knob up or down, how does the pipeline behave, and what does that mean for our customers and analysts?
  • Experience with AWS, Pulumi or SST.

Bonus points for experience in:

  • News data analysis (!), logistics, or energy
  • TypeScript
  • Web scraping (proxies, crawlers, etc.)
  • Structuring or maintaining enterprise-grade APIs

We don't expect you to arrive knowing everything. We do expect you to arrive with strong fundamentals, good instincts, and the drive to figure out what you don't yet know.

We care more about the right person than the right CV. A sharp, motivated junior and a seasoned engineer can both be the right fit for this role. What we're looking for is curiosity, ownership, and the ability to operate with minimal hand-holding. Compensation reflects experience.


What we offer


Full ownership from day one. You'll work directly with the founders. When you make a decision, it ships. You'll be a core part of how we expand into new areas, from the research phase through to production.

We go where the opportunity actually is. We have strong revenue and we're growing in the areas where we know we can win. We don't chase trends or move into spaces because someone else thinks we should. The roadmap is driven by what our customers need and what the data tells us.

A domain worth caring about. Our customers reroute shipments around typhoons, prevent factory shutdowns, and pull cargo out of war zones because we gave them the intelligence to act in time. The work you do will have consequences you can actually trace.

A founder who's done this before. One of our co-founders has previously scaled and exited a company. You'll be working closely with someone who knows what building something real actually looks like.

Great people deserve great conditions. Top-of-the-line MacBooks and screens, a proper office in the heart of Munich (Leopoldstrasse), Wellpass sports membership, competitive market-rate salary and VSOPs.


The kind of person who will thrive here


You've maintained something that breaks in production and you know what that actually feels like. You've built things without a detailed spec. You're comfortable sitting with an ambiguous problem, forming a hypothesis, and figuring out the answer before anyone tells you what to build. You're more energized by owning a problem end-to-end than by getting a ticket assigned. You'd rather work in a small team where your work is visible than disappear into a large org where it isn't.

You're probably not looking for your first job, and you want to build.


How to apply


Send a short email to careers@sea-sentinel.ai with:

  1. Two to three sentences on what drew you to this role specifically.
  2. One or two examples of work you've shipped. Tell us what problem you owned, what you built, what broke, how you knew whether it worked, and what you'd do differently. Links, GitHub repos, screenshots or a short walkthrough are all welcome. Please don't share confidential customer material. We'd much rather read a few rough sentences that are clearly yours than a polished page written by an AI. We care about what you have to say, not how polished it sounds. We use LLMs heavily every day, but we value engineers who built things before they existed. If you have a project from that time that you took from start to finish, show us.
  3. Your CV.


We’ll start getting back to applicants from November 24 onward.


Information on how we process your data: Privacy Notice for Applicants


Sea Sentinel AI is based in Munich, Germany. We work hybrid, and you will be required to be based in or around Munich. The start date is flexible, but our preference would be January 1.

Skills

Apply

See also

AI Engineering jobs by country — openings, pay and top skills →