Most maritime AI projects fail because they're built in labs, not operations centers. Models that work on clean datasets break on real supplier invoices. Agents that pass tests can't handle the exceptions that define maritime logistics. You'll build AI systems that operate in the messy reality of global shipping, extracting data from inconsistent PDFs, learning supplier-specific terminology, understanding the difference between "fuel pump" and "fuel injection pump" across six different part numbering systems.
What You'll Do
- Develop AI agents that automate maritime workflows: procurement request processing, quote comparison, shipment tracking, invoice reconciliation
- Build computer vision systems that extract structured data from PDFs, emails, and images (equipment specs, part numbers, invoices, delivery notes)
- Implement LLM-powered agents that understand maritime terminology, map between supplier systems, and learn from operator feedback
- Design and train models for entity extraction, document classification, and data validation specific to maritime operations
- Build the data pipelines and labeling workflows that enable continuous model improvement
- Integrate AI capabilities into product features, make models useful, not just accurate
- Develop evaluation frameworks that measure real-world performance, not just benchmark metrics
- Work with operators through our Ambassador Program, watch AI in production, understand failure modes, iterate based on actual usage
Who You Are
- 4+ years ML engineering experience with production deployments
- Strong hands-on experience with LLMs: fine-tuning, prompt engineering, agent frameworks, RAG systems
- Computer vision background: OCR, document parsing, layout analysis, text extraction
- Solid software engineering skills: Python, testing, deployment, monitoring
- Experience turning research into production: you know the difference between achieving good metrics and building reliable systems
- Comfortable with messy, unstructured data, you understand that real-world data doesn't look like benchmark datasets
- Pragmatic about AI: you know when to use a simple rule versus a complex model
- Customer-focused: you measure success by whether operators trust your AI, not just accuracy scores
- Comfortable working directly with operators and technical teams during deployments, you'll debug issues on-site and explain technical decisions in operational terms
Nice to Have
- Experience in document AI, information extraction, or NLP for enterprise applications
- Background in logistics, supply chain, or operations-heavy industries
- Familiarity with RPA frameworks or workflow automation
- Understanding of knowledge graphs, ontologies, or entity resolution
Why Join Narwhal
- Build AI that operators actually use, see your models in production immediately
- Work on genuinely hard problems: real maritime data is inconsistent, unstructured, and domain-specific
- Direct customer access, understand requirements from operators, not product specs
- Iterate fast based on real-world feedback, not lab experiments
- Equity stake in a company using AI to transform global trade
How to Apply
Send your resume and a note about why you're interested to careers@narwhal.ai
Tell us:
- What excites you about building for maritime operations
- A hard problem you've solved that's relevant to this role
- Why you want to join at this stage
We'll respond within 48 hours.
