Applied ML/AI Engineer

About the role

We are looking for a product-minded Applied ML / AI Engineer who connects data, models, software, and business outcomes. You will evaluate and apply AI to real product and operational challenges in maritime crewing, taking ambiguous problems from discovery and prototyping through production deployment and continuous improvement.

You will own the quality and practical application of AI: selecting the right models and techniques, designing data and evaluation loops, and turning uncertain model outputs into dependable product capabilities. You will also advise product, operations, engineering, and management teams on where AI is genuinely useful, what risks and limitations to expect, and how to introduce automation without unnecessary cost or complexity.

This is applied engineering, not foundation-model research. You will work with the AI Software Engineer, product stakeholders, crewing operators, and QA to build systems that improve real business outcomes.

A key part of the role is choosing the right level of intelligence for each problem: deterministic rules, classical ML, specialized models, LLMs, agentic workflows, or human review.

What you will build and own

• Document intelligence for CVs, licenses, certificates, IDs, employment records, and photos of paper documents.
• OCR and multilingual extraction into strict schemas with per-field confidence.
• Document classification, expiration checks, missing-field detection, consistency checks, and source-priority rules.
• CV and employment-history parsing, deduplication, conflict resolution, and profile-quality scoring.
• Matching and ranking models for seafarers and vacancies, with useful explanations for operators and customers.
• AI-assisted scoring and verification workflows with fairness and reliability checks.
• Speech and conversation intelligence for AI calls and seafarer communication: transcription, intent and field extraction, summaries, and safe next actions.
• AI-driven follow-up workflows that can contact seafarers, request missing or updated information, clarify answers, and help keep profiles current.
• Business logic for deciding when an AI system may act automatically and when it must ask for confirmation or involve an operator.
• Human-in-the-loop feedback loops that turn operator corrections into curated evaluation data.
• Golden corpora, regression suites, model comparisons, confidence thresholds, and evaluation metrics.
• Model, prompt, schema, and dataset versioning with safe rollout and rollback practices.
• Monitoring of accuracy, latency, cost, fallback rate, correction rate, reliability, and business impact.
• Practical guidance on when to use an LLM, deterministic rules, classical ML, a specialized model, or a human review step.
• AI safety and privacy practices for sensitive maritime data: data minimization, consent-aware workflows, safe tool use, output validation, auditability, and escalation of uncertain or high-impact cases.
• How you will work
• Participate in discovery sessions with product, operations, engineering, QA, and crewing stakeholders. Turn ambiguous business problems into realistic technical scopes, measurable hypotheses, and working proof-of-concepts. Demonstrate prototypes to stakeholder teams and use operator feedback to improve data, workflows, and model behavior. Design and run structured experiments, then move validated solutions into production and support their ongoing improvement. Reason about trade-offs between accuracy, latency, cost, reliability, privacy, safety, and engineering complexity. Explain model behavior, limitations, and recommendations clearly to technical and non-technical audiences. Provide practical AI consultation and knowledge transfer across product, operations, engineering, and management.
• What we are looking for 4+ years of applied ML, NLP, data science, or AI engineering experience in production systems. At least one ML- or LLM-powered capability shipped to real users or operators. Strong understanding of document AI, OCR, information extraction, classification, ranking, or conversational AI. Practical experience with structured outputs, tool calling, embeddings and retrieval, model routing, and evaluation pipelines. Strong Python skills and comfort integrating models into APIs and production services. Experience creating labeled datasets, golden corpora, quality metrics, error taxonomies, and regression tests. Ability to reason about noisy data, confidence, false positives and negatives, source conflicts, and human review. Understanding of privacy, PII, consent, prompt injection, and safe handling of sensitive user data. Ability to evaluate trade-offs between accuracy, latency, cost, reliability, privacy, and system complexity. Ability to explain model behavior and translate business problems into realistic AI solutions. Comfortable taking ownership of ambiguous problems from discovery through production and ongoing support. Working English, written and spoken.
• Nice to have Multilingual NLP and robustness to low-quality documents and photos. Speech-to-text, telephony, voice agents, or call-center automation. Recruiting, HR technology, maritime, logistics, or marketplace experience. Experience with AWS, S3, vector databases, model observability, or MLOps. Experience with fairness, exploitability, or verifiable credentials. Experience using AI coding agents for experiments or production implementation.
• Location and work format Remote work may be considered for an exceptional candidate. Apply with a short note about the most interesting system you’ve built and a link to code or a write-up.