Maersk Tankers – AI & data science collaboration for a global maritime logistics leader

Tatia Tatoshvili

Last updated on August 12, 2025 | 17 min read

Introduction

Solving complex logistical challenges in the global maritime industry requires more than operational know-how, it demands advanced analytics, scalable data infrastructure, and precision innovation. From optimizing vessel fuel efficiency and minimizing port turnaround times to improving route planning based on AIS data, modern maritime logistics hinges on data-driven decision-making at scale.

Maersk Tankers, a leading global operator of tanker vessels, partnered with DevsData LLC on a six-month AI and data science initiative designed to explore opportunities for optimization across various layers of their maritime logistics operations.

The goal: To leverage modern data science techniques in support of optimization, performance, and strategic insight, while ensuring confidentiality of all domain-specific knowledge.

This case study outlines the collaborative approach, team structure, and delivery model, demonstrating how DevsData LLC embedded a high-calibre technical team into a strategically vital project for one of the world’s most established maritime logistics brands.

Client overview

Maersk Tankers is a global leader in commercial tanker operations, managing one of the world’s largest fleets of product tankers. With a long history in the maritime industry, the company blends decades of seafaring expertise with a forward-looking commitment to digital transformation, sustainability, and data-driven optimization.

Headquartered in Denmark, Maersk Tankers operates globally and services a broad portfolio of clients in the energy and transport sectors. Their work involves dynamic fleet allocation, fuel performance analysis, operational routing, and real-time commercial decision-making, all of which rely on complex data ecosystems, including satellite AIS feeds, weather forecasts, fuel consumption logs, cargo load data, and port congestion metrics. These data streams are critical for optimizing voyage planning, reducing emissions, and maximizing commercial outcomes.

As part of their ongoing innovation strategy, Maersk Tankers partnered with DevsData LLC to explore how AI and data science could unlock further operational efficiencies, enhance predictive accuracy, and support long-term commercial objectives by minimizing variability in voyage performance and enabling more informed decision-making.

At a glance: Maersk Tankers

• Team size: +100000 employees
• LinkedIn followers: 152K
• Headquarters: Copenhagen, Denmark

Maersk is leading the charge with a bold commitment to e‑methanol, the world’s first commercial-scale, low-emission marine fuel produced at the €150 million Kasso plant in southern Denmark. This facility uses renewable energy and captured CO₂ to produce 42000 tonnes (≈53 million L) per year, enough to fuel one 16000‑TEU container ship on the long-haul Asia-Europe route.

Challenge

To maintain its competitive edge in an increasingly data-centric shipping industry, Maersk Tankers set out to deepen its use of advanced analytics.

The goal wasn’t just incremental improvement, it was to reimagine how real-time operational data could power smarter, faster decisions across vessel routing, fuel usage, and logistics coordination. This required going beyond conventional tools and engaging AI and data science at a strategic level.

However, several challenges emerged:

• Domain-specific complexity

  Maritime logistics involves numerous variables, weather conditions, vessel performance data, and port regulations which are difficult to model using generic data science techniques. For instance, accurately predicting port congestion requires integrating satellite AIS data, berth availability, historical docking patterns, and regional regulations. Maersk Tankers needed experts who could navigate these complexities and build tailored, production-grade AI solutions for real-time decision-making.

• High standards for technical credibility

  Given the strategic importance of the project, Maersk Tankers required professionals with strong theoretical backgrounds, hands-on experience in predictive modeling, and the ability to collaborate with internal technical and commercial teams.

• Rapid execution within a fixed 6-month engagement window

  The project required swift onboarding and meaningful progress within a short 6-month timeframe, with minimal room for iteration delays or misalignment. It was critical that any external team could operate at the same level of rigour as Maersk’s in-house experts.

• Sensitive, confidential environment

  Given the strategic nature of the project, all work was conducted under strict confidentiality. Since DevsData LLC operated through an embedded delivery model, augmenting Maersk’s internal teams with external specialists, it was crucial that contributors adhered to the company’s data governance policies. To ensure compliance and protect sensitive operational data, the team worked exclusively within Maersk’s secure, access-controlled cloud environments. No data was stored locally, and version control protocols, code reviews, and documentation followed Maersk’s internal standards. This approach ensured full alignment with internal security frameworks while maintaining delivery velocity.

DevsData LLC’s approach

DevsData LLC was selected for its deep technical expertise, proven track record in high-stakes data science initiatives, and its ability to operate effectively within domain-specific, confidential contexts. Our approach with Maersk Tankers was shaped by prior engagements in similarly complex, data-intensive environments.

In a long-term collaboration with Kroll, Inc., we delivered scalable data pipelines and backend systems designed to process high-volume financial intelligence, insights from which directly informed our architectural and security decisions for Maersk’s maritime models. Additionally, our experience supporting the Norwegian SaaS company Memory AS with senior AI-focused engineering talent across Europe and Asia enabled us to rapidly assemble and integrate a high-caliber team, ensuring immediate value delivery in Maersk’s fast-paced setting.

Our broader track record across the Nordic region further reinforced our domain familiarity and adaptability. We’ve supported clients such as Memcare AS, Kyoto AS, and Varner AS, as well as Circle K’s Swedish headquarters, providing tailored engineering solutions and strategic recruitment across multiple verticals.

To meet Maersk Tankers’ expectations, we assembled a focused, senior-level delivery team:

• 3 Senior Data Scientists, each with 7-10 years of experience in applied machine learning, optimization, and signal processing. The team brought domain expertise from transportation, energy, and financial forecasting, allowing them to quickly adapt to the maritime logistics context and deliver high-impact solutions with minimal onboarding.
• 1 Full Stack Developer with strong backend and data infrastructure capabilities, responsible for building internal tooling, data visualisation layers, and integrations with Maersk’s digital environment.

Key features of our engagement included:

• Rapid onboarding and client-side integration

  Our specialists were embedded into Maersk’s cross-functional team within days, participating in daily stand-ups, sprint planning, and documentation workflows. The engagement followed Agile methodology, with collaboration managed through Jira for issue tracking and Confluence for shared documentation. All team members operated under Maersk’s internal security and access protocols, ensuring seamless integration and compliance from day one.

• Tailored, domain-aware modeling

  The data scientists collaborated closely with Maersk’s operations and analytics stakeholders to create custom models and tools that could support specific use cases. This included iterative hypothesis testing, rapid prototyping, and validation using historical data.

• All progress logged via version-controlled Git repos and Jira

  All code, research findings, and tools were delivered via secure, client-approved platforms. Progress was documented through bi-weekly reports, and all models were designed to be fully explainable and auditable for internal reviews.

• Consultative partnership

  DevsData LLC provided more than just technical execution. We worked as a strategic partner, raising design concerns, proposing data acquisition improvements, and advising on long-term maintainability of the models.

Solution

To meet Maersk Tankers’ need for sophisticated data analysis within a secure, high-stakes environment, DevsData LLC implemented a focused and modular solution designed to support data-driven decision-making across key operational areas, without compromising strategic confidentiality or internal workflows.

Our delivery model focused on four pillars:

1. Iterative modeling and domain-tailored analytics

   We developed custom models tailored to the unique challenges of maritime logistics, capable of handling noisy, incomplete, and volatile data drawn from operational and environmental sources. These models integrated a blend of time-series forecasting, optimization algorithms, and operational data to support better routing, scheduling, and fuel efficiency decisions.

   We employed a lean experimentation framework, enabling Maersk Tankers to test, validate, and deploy data-driven insights in sprints, minimising the risks of full-scale rollout while improving iteration velocity.

2. Secure, embedded collaboration

   All collaboration occurred within Maersk Tankers’ internal environments, ensuring that proprietary data, modeling outputs, and insights remained fully under the company’s control. Our team adapted to internal documentation styles, DevOps protocols, and versioning standards to maintain seamless integration.

   The full stack engineer focused on building internal dashboards, data pipelines, and visualization tools that allowed Maersk Tankers’ analysts and managers to interpret model outputs easily and apply them in daily operational decisions.

3. Production-ready data science tooling

   Rather than building standalone prototypes, our team delivered maintainable, production-level components that could be extended or owned by Maersk Tankers post-engagement. This included:

   • Modular Python codebases, tested and versioned
   • Custom notebooks for exploratory analysis
   • Deployment-ready scripts for model execution
   • Visualization layers for cross-functional teams, built using tools like Plotly Dash, Power BI, and Looker

   All solutions were compatible with Maersk’s cloud and on-prem environments, respecting performance, security, and compliance requirements.

4. Stakeholder enablement and knowledge transfer

   From the outset, our work included documentation, walkthroughs, and Q&A sessions to ensure long-term usability by internal teams. This ensured that Maersk Tankers was not dependent on external consultants to maintain or evolve the delivered solutions.

   We also provided insights on model monitoring, retraining strategies, and ways to scale the approach for broader adoption within the organization.

   Parameter	Detail
   Duration	6 months
   Team Composition	3 Senior Data Scientists, 1 Full Stack Developer
   Focus Areas	Data modeling, optimisation, internal tooling
   Tools & Tech	Python, Jupyter, cloud integrations, dashboards
   Delivery Style	Embedded, secure, sprint-based iterations
   Result	Modular, reusable analytics assets built for Maersk Tankers