Explore our industry expertise
Success stories
U.S. car manufacturer accelerates innovation with AI platform

U.S. car manufacturer accelerates innovation with AI platform

AI platform to efficiently process vast amounts of data

U.S. car manufacturer has embraced AI as a cornerstone of its innovation strategy. To support its ambitious goals, the manufacturer partnered with GFT to develop a cloud-native AI-driven data platform designed to power advanced driver assistance systems (ADAS).

A futuristic autonomous car levitates above a digital platform, with AI, biometric, and connectivity icons highlighting advanced technology integration in automotive innovation. — Discover how a leading car manufacturer is leveraging AI-powered platforms to revolutionise the automotive industry, enhancing efficiency, safety, and connectivity.

Rapid development and scalability

GFT developed an AI data platform in just seven months, scaling it to production capacity in two additional months.

Global data collaboration

Enables worldwide data access and AI tool integration for seamless teamwork.

Enhanced productivity

Empowers car manufacturer's data scientists to collaborate, analyze data and build pipelines with Vertex AI Notebooks, enhancing efficiency.

Challenge

Addressing the challenge of data management

As the automotive’s ADAS capabilities grew, so did the volume and complexity of the data required to train and refine its Machine Learning (ML) models. The U.S. car manufacturer needed a solution that could efficiently manage, process and analyze vast, petabyte-sized (1000 terabytes) datasets from various sources, including cameras, sensors and other connected devices.

GFT’s expertise in AI and data management proved invaluable in addressing the challenges identified by the company. By building a hybrid data ingestion and AI platform in partnership with Google Cloud, GFT helped create a centralized data repository in streamlining workflows and improving data quality.

Strategic platform for innovation

The AI data platform is more than just a data storage solution, rather it is a powerful platform that enables the U.S. car manufacturer’s data scientists to collaborate, experiment and innovate. By providing a unified environment for data exploration, labeling and integration with data discovery and security controls, the data platform accelerated the development of new ADAS features and capabilities.

Beyond ADAS: expanding use cases

While ADAS was the initial focus, the car manufacturer is now exploring additional applications for the AI data platform, including:

Visual Inspection for Manufacturing: Identifying defects and quality issues in real-time;
Vehicle Quality Assurance: Ensuring vehicles meet stringent standards prior to delivery;
Servicing and Warranty Management: Optimizing maintenance processes and reducing costs.
Warranty Claims Management: Reducing warranty costs by automatically auditing the claims to eliminate erroneous submissions

Foundation for future growth

The data platform has become a strategic asset for the U.S. car manufacturer, providing a flexible and scalable foundation for future AI initiatives. As the company continues to push the boundaries of automotive technology, the data platform will play an important role in driving innovation and delivering an exceptional customer experience.

"The data platform offers our client an efficient solution for managing unstructured data in AI and Machine Learning, with powerful tools for ingestion, processing and more. It is a leap forward in data-centric AI development."

Brandon Speweik

Head of Industry Sales and Strategy

GFT USA

Portrait of Brandon Speweik wearing a dark blazer and white shirt, smiling confidently. — Connect with Brandon Speweik at GFT.

Engagement

Bringing an AI vision to life

GFT assembled a multidisciplinary team to scope the project, evaluate options and validate software components through a proof of concept. GFT’s team was comprised of architects, data engineers, DevOps engineers and AI specialists.

At the project’s outset, GFT and the U.S. car manufacturer agreed on key performance indicators to measure the progress and success of the project against unambiguous criteria. The approach also facilitated periodic review and user feedback to keep the project on schedule with its high-level objectives.

The project centered on the car manufacturer’s AI visual data management platform. Created by GFT and the U.S. car manufacturer, it is a bespoke platform for managing image and video data for ML projects, while also managing indexes of data to streamline ML engineering workflows and ensure that they are more efficient. In practice, it powers an integrated ecosystem that enables application teams to quickly and easily create golden datasets and utilizing them seamlessly across multiple applications and platforms. This eliminates duplication of data and leads to significant cost savings. The data platform is also integrated with MDK (Machine Learning Development Kit) – the company’s custom ML solution built with a variety of Google services.

To optimize performance and cost, the U.S. car manufacturer’s data platform harnesses the power of multiple Google Cloud Platform services. Examples are Cloud Run, Vertex AI, Big Query and Cloud Storage. Additionally, it uses the vehicle manufacturer’s high-performance computing platform.

Automated pipeline for deploying required Google Cloud services for model maintenance:

Google Cloud Storage Repository for dataset storage and to enable training;
Web application for data and image labeling;
Model training services using generated datasets;
Model evaluation applications for accuracy and performance;

The new hybrid model simplifies integration with other systems and workstreams. The completed platform comprises four complementary modules:

Extractors: AI Models that are built into the AI data platform. They assist in extracting additional data from images, video, and audio sources such as similarities, metadata, and PII for de-identification;
Golden Data: A web application for working, querying and searching datasets –namely, images and videos – then labeling and connecting with other applications;
Data Pipelines: Ad hoc automated flows for each use case, facilitating the sending and monitoring of sensor data;
Search Visualization: Tools for working with images and datasets.

Benefit

A secure, versatile platform that drives AI uptake

GFT built a prototype platform within seven months and over the subsequent two months, the platform was scaled to production capacity. Using its proven global delivery model, GFT continuously enriched product functionality in response to feedback. Currently, the data scientists of the U.S. car manufacturer can browse, analyze and easily access a rich source of quality data, which can be utilized in building pipelines with either a user interface or software developer kit.

The platform places data at the heart of the company’s manufacturing process. Initially commissioned to power the ADAS, it is now evaluating many new use cases, including AI visual inspection for manufacturing, vehicle quality assurance and servicing and warranty management. As a cloud-native repository, the data platform has solved the company’s challenge of decentralized information.

The new platform is flexible enough to ingest data from new and disparate sources, such as in-car driver information for analysis and comparison with other datasets, empowering the U.S. car manufacturer to derive more value from its data with minimal investment. With a single source of quality data, the data scientists are now more creative and productive.

Data can be analyzed within the platform and accessed by the car manufacturer’s data scientists in locations worldwide, encouraging collaboration and uptake. The data platform’s users can also use other third-party tools integrated into the platform, which empowers them to work with data, for example, using Vertex AI Notebooks.

Although the platform is live, GFT continues to add new functionality to support new uses cases and data sources. The data platform has become a benchmark within the U.S. car manufacturer, as ML is driving efficiency in many business processes. The data is also being integrated with third-party ML applications for use across various platforms.