Projects

Description: In the intricate world of port operations, safety and efficiency reign supreme. We developed an AI-powered solution for a marine terminal, processing live video feeds to identify potential hazards and immediately notify relevant personnel, ensuring top-notch safety standards.

Location: Romania

Industry: Logistics

Solution: ML, Data Analytics

Engagement model: Solution Based

Key Technologies: CCTV, Python,PyTorch, Streamlit, MongoDB

Challenge:

In the complex and often hazardous environment of port operations, ensuring safety is paramount. With this in mind, our data team embarked on a mission to develop a Proof of Concept (PoC) that would automate port operations and significantly heighten safety levels.

This endeavor aimed to process real-time video feeds, employing Machine Learning (ML) models to keenly analyze the positioning and movement of individuals, thereby promptly notifying relevant personnel of potential hazards.

Solution:

The infrastructure for this innovative solution harnesses CCTV cameras, which continually relay video feeds via FFMPEG. The proprietary ML models developed by us, are meticulously designed to detect not just the presence of an individual, but also their exact positioning, movements, and even safety gear. The system can identify if an individual is approaching a danger or hazardous zone and check whether they are donning safety vests or helmets, crucial attire in such environments.

Beyond immediate safety measures, the system also offers extended analytics, like evaluating the time a truck spends within the port terminal. This insightful data can be instrumental in comprehending challenges associated with loading operations and highlighting areas ripe for improvement.

Results:

The result is a pioneering application tailored for the modern port, aiming to automate crucial aspects of inland operations while placing a premium on safety. With its real-time monitoring and immediate alert system, the risks associated with cargo lifting, heavy machinery movement, and other hazardous tasks are substantially mitigated. 

By ensuring operators and staff are promptly alerted when entering risky zones without adequate safety gear, the application promotes a culture of safety and precaution. Our Smart Marine Terminal application is a testament to how cutting-edge technology can be leveraged to create safer, more efficient working environments.

Location: UK, Hong Kong

Industry: Logistics

Solution: ChatBot

Engagement model: Solution Based

Key Technologies: C#, .Net, Azure AI Components

Description: To accentuate user interactions for a Global Forwarding company, we harnessed the capabilities of Azure AI Components to develop a responsive chatbot. This AI-driven assistant streamlined user queries, enhancing their experience on the company's tracking and billing application.

Challenge:

A leading Global Forwarding company approached us with a distinct need: the creation of an intelligent chatbot focused on self-service and exceptional graphic design. With the ever-evolving digital landscape, it was imperative for the client to provide users with an intuitive and efficient means of accessing information without being overwhelmed.

Solution:

Leveraging the advanced capabilities of Azure AI Components, our team undertook the task of developing the chatbot. Our machine learning engineer meticulously trained the chatbot, familiarizing it with potential user queries to ensure the most accurate and relevant responses were delivered. This training ensures that the chatbot is adept at understanding user needs and provides instantaneous, accurate answers, making user interactions smooth and productive.

Results:

The introduction of the ChatBot marks a significant enhancement to the company's tracking and billing application. Serving as an interactive guide, the chatbot facilitates users in effortlessly navigating the platform, accessing vital information, and ensuring they aren't overwhelmed by extensive data sets. By prioritizing self-service, the chatbot fosters user independence and confidence, streamlining their experience and ensuring the platform is user-friendly and efficient. This innovation underscores our commitment to leveraging cutting-edge technology to deliver solutions that enhance user experiences and business operations.

Description: Drowning in extensive video transcripts from marathon meetings, a Global Forwarding company required an efficient search solution. We innovated by leveraging Elastic Search combined with machine learning, ensuring rapid and precise pinpointing of key topics within the transcripts, saving precious time and effort.

Location: UK, Hong Kong

Industry: Logistics

Solution: AI, Elastic search

Engagement model: Solution Based

Key Technologies: AI, Elastic search, Apache, React TypeScript, Google Bucket

Challenge:

The leadership team of the Global Forwarding company often found themselves in prolonged meetings, generating extensive video transcripts. The sheer volume of these transcripts posed a challenge when attempting to search for specific subjects or topics of discussion. The company sought an efficient mechanism to swiftly locate key subjects within the video transcripts using elastic search.

Solution:

In response, we devised a state-of-the-art tool, developed prior to the release of ChatGPT. This tool first analyzes video content to generate a comprehensive transcript. Leveraging the power of Elastic Search combined with machine learning, the tool was trained to swiftly identify and pinpoint subjects related to specific topics within these transcripts. This integration of Elastic Search with ML capabilities ensured that the search process was both dynamic and accurate.

Results:

The outcome was a transformative tool that significantly minimized the time spent searching through extensive transcripts. With the trained machine learning algorithms at its core, the tool rapidly and efficiently locates the necessary information, allowing management to quickly reference specific discussions and decisions. This solution not only optimized the company's approach to information retrieval but also showcased the potential of integrating Elastic Search with machine learning in streamlining business processes.

Description:

Our project with a server administration company in the UK involved addressing a critical issue: the lack of comprehensive documentation following the departure of a senior developer. This gap posed significant challenges in understanding the existing system architecture and functionality.

Location: UK

Industry: IT, Global Forwarding

Solution: Integration, Data Manipulation, Data Load

Engagement model: Staff augmentation

Key Technologies: Talend, C#, SQL

Challenge:

The primary challenge was to fill the documentation void to ensure continuity and clarity in the server administration processes. This task was crucial for maintaining and enhancing the system's efficiency and reliability.

Solution:

We approached this challenge by integrating AI technology to decipher the system's architecture and functionality. This innovative solution facilitated a deeper understanding of the existing systems without relying on legacy documentation. Additionally, we implemented a server notification system, seamlessly integrated with Slack and email, which could be configured by the admin or through the platform.

This system was designed to provide real-time alerts and updates, enhancing the responsiveness of the server administration team. Another significant aspect of the project was the Windows installation over a network using PXE, streamlining the deployment process. Lastly, we conducted a thorough study and analysis of an open-source billing application, laying the groundwork for its future integration into the company’s systems.

Result:

The integration of AI for analysis and documentation, along with the implementation of the notification system and network-based Windows installation, significantly improved the company's server administration processes.

Our solution, developed using Talend, C#, and SQL, under the staff augmentation engagement model, enabled the company to overcome the challenges posed by the absence of adequate documentation and modernize their server management approach. The preparatory work for the billing app's integration further set the stage for enhanced operational capabilities in the future.

Technical Details:

• Photogrammetry and structure from motion algorithms for 3D model creation


• Convolutional Neural Networks (CNN) for object classification


• Pixel pattern matching and opacity variance detection


• Over 1,000 categories for industrial equipment identification

Challenge:

Manually identifying industrial objects, especially when they are damaged or obscured, was slow and prone to errors. An automated solution was needed for faster, more reliable identification.

Solution:

Using machine learning and advanced image analysis, we developed a solution that builds 3D models from multiple angles and classifies objects based on distinguishing features like shape and texture.

Results

The solution achieved 7% higher classification accuracy compared to manual methods, significantly improving equipment audits and inventory management while reducing misidentification events.

Description:
Our team developed an AI-powered data pipeline quality assurance system for a client with complex data engineering needs. The system continuously validates extraction, transformation, and loading (ETL) processes across various databases and data lakes, ensuring data integrity and operational reliability.

Technical Details:

• AI-powered validation of ETL processes


• Real-time log analysis and statistical data integrity checks


• End-to-end data lineage tracking


• Alerts for deviation detection and remediation


• Actionable debugging information for engineers

Challenge:

The client faced significant challenges in maintaining data integrity across massive and disparate data sources. The complex nature of their data pipelines made it difficult to identify and resolve data issues before they impacted downstream analytics.

Solution:

We developed an AI-driven quality assurance system that continuously monitors the client’s ETL processes, performing real-time log analysis and statistical integrity checks. The system tracks data lineage from source to destination and provides immediate alerts when deviations occur, allowing engineers to quickly identify and correct issues.

Results

Within months of deployment, the system identified over 50 pipeline failure points, preventing disruptions to downstream analytics. The solution has since become an essential tool for the client’s data engineering team, enabling faster development of new data migration routes and ensuring robust data integrity.