Success Stories - TantusData

AI-Powered Legal Chatbot

Paulina Wiśniak — Mon, 25 Nov 2024 12:45:05 +0000

Client

A startup specializing in providing global access to legal documents for clients. They aim to develop an innovative chatbot that connects with their knowledge base, creating a new way for users to interact with legal information.

Opportunity

This solution offers professionals who work with legal documents a convenient and reliable way to ask an AI system for answers, significantly speeding up legal research. The chatbot is designed to respond quickly, be intuitive, and provide not only an answer but also AI-driven reliability indicators and links to supporting documents. This approach enables lawyers and paralegals to access ready-to-use answers, with only the final step of verifying the referenced documents.

Delivery

Working closely with the customer, we began by exploring ways to connect their extensive legal document database with AI capabilities for answering legal questions. We analysed the performance of various large language models in responding to legal inquiries and identified how the documents they referenced could be linked back to the customer’s vast database.

We also developed a verification mechanism for assessing the accuracy of responses and designed an intuitive display for presenting results. Given the commercial nature of the product and the need for rapid market validation, we integrated the chatbot with a payment provider and implemented user registration and account management features.

Effect

We developed the solution from scratch to a fully functional, end-to-end version. It enables users to focus directly on the content of the answer and the specific document sections they need, eliminating the tedious work of locating and reviewing documents. This approach has accelerated legal research by up to 500%.

The post AI-Powered Legal Chatbot appeared first on TantusData.

Customer support on AI steroids

Paulina Wiśniak — Thu, 14 Nov 2024 12:33:13 +0000

Client

The large enterprise leader in the telecommunications industry across the Nordic-Baltic region. Aiming at improving efficiency of their customer support by introducing essential AI infrastructure and services.

Opportunity

Providing customer support representatives with relevant information, and AI generated proposal increases the effectiveness of the Team as well as customer satisfaction. It is also a great first step for enabling more features, including more advanced suggestions for the support representatives as well as fully automating some of the requests.

Delivery

Together with the client’s team we started by identifying already known common challenges tackled by customer service. We identified types of customer requests but also what part of the process could be improved by AI in order to increase efficiency.

We did identify the low hanging fruits by talking to customer service representatives and analysing the historical data including emails and voice calls.

Effect

The solution was introduced from zero to a first working end to end version. It increased customer satisfaction by 20% and significantly increases customer support awareness of the upsell opportunity. At the same time the data collected from the first version will be analysed as an input for the future improvements.

The post Customer support on AI steroids appeared first on TantusData.

Introducing Databricks for a fintech startup

Magdalena Majka — Fri, 26 Apr 2024 06:32:54 +0000

Client

Tradeteq is a fintech startup based in the UK.

Opportunity

The Databricks platform is more cost-efficient and offers more flexible scaling options than an SQL database (including cloud offerings). Keeping the SQL solution would ramp up the infrastructure cost by an order of magnitude, while the Databricks & Spark platform scales linearly and is significantly more cost-efficient in our client’s case.

Delivery

As a startup, the company focuses on short delivery time, quick feedback loop, and transparency regarding the current state of features in progress. TantusData has successfully accommodated this approach and delivered accordingly. Within a month and a half, all major data jobs were migrated from Azure Functions applications to Apache Spark jobs. The process includes onboarding and familiarising the client’s engineers with the solution.

Effect

The implemented data platform resolved the challenge of scaling the existing MSSQL-backed batch processing. The platform is ready for a 100x increase in events without affecting the transactional backend system. It not only enables future growth but is significantly cheaper than the previous solution.

The post Introducing Databricks for a fintech startup appeared first on TantusData.

Customising Databricks for Comprehensive Business Metrics and Data Integration

Magdalena Majka — Thu, 18 Apr 2024 13:20:20 +0000

Client

A startup that is providing software for restaurants and restaurant chains to streamline managing them.

Opportunity

The platform integrated all different data sources, giving a full view of sales that did not require manual computations. The business can now make data-driven decisions instead of relying solely on intuition and partial data samples.
A common platform with all the data in a single place is also an enabler for Data Science and Machine Learning activities.

Delivery

The delivery process included not only creating a framework to compute generic APIs but also working on ensuring data quality. TantusData has a flexible approach to all projects and quickly adapts to changing requirements or priorities, which are common in startup environments.

Effect

The TantusData team created a complete solution ingesting data from the raw, unprocessed sources (APIs, manual data, operational databases) up to the final visualization layer for the end users.

The post Customising Databricks for Comprehensive Business Metrics and Data Integration appeared first on TantusData.

Introducing Apache Airflow

Magdalena Majka — Thu, 08 Feb 2024 15:10:08 +0000

Client

An international telecommunication company based in Austria.

Opportunity

Using a well-known and supported tool instead of an in-house developed tool can free developers to let them work on other areas instead of reinventing the well-available wheel.
Ability to onboard more people into a standard tool – there is UI and documentation, which allows not only data engineers to understand the pipeline logic.
Hiring and introducing new people to a project using a well-known toolkit is easier and faster.

Delivery

The migration began with working with the infrastructure team to deploy the Apache Airflow tool, following the best practices already in place. TantusData delivered scripts to recreate the deployment and ensured that sysops would be familiar with the tool and understand what and how it is deployed and configured and what security concerns must be addressed.
Having a deployed tool in place, TantusData began working with the data team to migrate a part of the existing pipeline. It is to rewrite a current part of the code and use it as an example of migrating other parts. The data team also had a chance to try using the tool on their own.
The PoC was concluded by showcasing monitoring options for the tool to ease up daily maintenance. TantusData also presented recommended next steps in terms of tips for the development team and the administration and maintenance.

Effect

The client was able to evaluate the Airflow tool, how it fits their platform and what they would need to do to integrate it.
The migration plan has brought up a discussion on reviewing and refactoring the data pipeline. Being able to see a visualisation of all tasks gives everyone a broad view and encourages discussing engineering matters like the manageability of graphs of over 100 nodes.

The post Introducing Apache Airflow appeared first on TantusData.

Hybrid Cloud Data Platform: Merging On-premises and Cloud Analytics.

Magdalena Majka — Fri, 15 Sep 2023 12:32:09 +0000

Client

A leading data and analytics division of a major telecommunications firm in Europe.

Opportunity

Advanced Data Analytics stands as a cornerstone for large telecom entities. However, its effectiveness is contingent not just on possessing data, tools, and skilled personnel, but on ensuring that these tools and processes adhere to stringent security and data governance benchmarks. Such standards fortify the organisation’s credibility, appeasing both clients and investors, while also mitigating potential regulatory penalties.
Navigating the realms of hybrid Cloud architecture amplifies the importance of data governance and security. Determining which datasets are processed within the public cloud versus those retained in-house becomes crucial to prevent any data breaches or mishandlings.

Delivery

The project’s genesis revolved around ameliorating the inherent vulnerabilities within the Cloudera platform. With the foundational security measures in place, the transition towards a hybrid cloud solution ensued. The AWS platform, renowned for its scalability and suite of advanced tools, was integrated, yet a fraction of critical datasets was deliberately retained in-house.

Effect

Post-implementation, Data Scientists were ushered into an environment that was both scalable and secure. Freed from the shackles of prior technical glitches, they could channel their expertise into leveraging more contemporary tools. They were also assured of consistent processing resources for their daily activities. This efficiency prompted the enterprise to integrate more Data Scientists and analysts into the platform, empowering them to distill genuine value from the vast swathes of data.

The post Hybrid Cloud Data Platform: Merging On-premises and Cloud Analytics. appeared first on TantusData.

Urgent manpower provision.

Magdalena Majka — Thu, 02 Feb 2023 12:04:36 +0000

Client

One of the main telecommunication service providers in the Nordic and Baltic areas. It owns a leading IPTV distribution platform in Sweden with video-on-demand and over-the-top streaming service that can be accessed across the EU.

Opportunity

The team, which our consultant has joined, is responsible for providing insights and reports about the company’s platform customers’ digital usage. These are fuelling all decisions regarding the purchase of content from external providers and experience. It also provides managerial insights into market penetration and effectiveness. Timely data and solutions provided by this team are imperative for the company’s prosperity.

Delivery

Our consultant was quickly onboarded and worked as a regular part of the team, helping deliver day to day activities and new features. These included: ongoing support for existing data pipelines, optimising existing data pipelines, onboarding new data sources into the common data layer, extending and creating new dashboards to provide business users with in-depth insights on the customer base. The development process involved identifying bugs in the core data and fixing issues for more reliable results. Other fixes were needed to solve the issues of undocumented codebases, multiple dependencies and tangled business logic.

Effect

By taking ownership of dedicated tasks, our consultant was able to improve several solutions based on his expertise. All tasks were completed by the deadlines. Projects delivered, work completely catched up. Giving the company ample time to hire new members, who meet their criteria.

The post Urgent manpower provision. appeared first on TantusData.

Sharing the magician’s toolkit.

Magdalena Majka — Thu, 02 Feb 2023 11:51:01 +0000

Client

A nordic telecom company. One of the leading global contenders.

Opportunity

Apache Spark fails affected the production pipeline, which was a part of data delivery engagement. Failure to deliver these data on time means missed SLA. Subsequently, fines, angry customers and reputation issues. On top of that a permanent fix would save time spent on restarting failed data jobs. It would reduce maintenance, provide better reliability and the issue of failures occurring over the weekends would also be solved.

Delivery

We identified the problem as a performance issue with Apache Spark job caused by incorrect partitioning configuration. We fixed the job, keeping in mind that the root cause of it is something that can be prevented with the appropriate know-how. Therefore, we made sure that we are transparent about the case and our work on fixing the issue. We led a knowledge sharing session and created a KB article for the maintenance team. It was to ensure that they would be able to fix such issues in other jobs by themselves and better understand the applications they own.

Our session covered:

A brief explanation about Spark partitioning – how to tune memory and how it relates to the job itself.
A description on how the issue happened – what exactly Spark did on the cluster and why.
A short note on how to mitigate such problems in the future and how to anticipate those during development; including ‘future reading’ with more elaborate articles on the topic.

Effect

We have fixed the issue, thus ensuring no more missed SLA as well as saving manual labour and other maintenance costs. We ensured that other occurrences of such issues will not have a negative impact by sharing the relevant knowledge with the team.

The post Sharing the magician’s toolkit. appeared first on TantusData.

Towering over Telcos.

Magdalena Majka — Thu, 02 Feb 2023 11:45:58 +0000

Client

A leading telecommunications company, media house and ICT provider in the Nordic and Baltic regions. They enable digital societies we live in with essential digital infrastructure and digital services.

Opportunity

The issues not only obstructed day to day work causing repetitive, redundant tasks and providing unreliable data. They delayed development, making it hard to capitalise the inputs. There were also other resulting losses, because the velocity of issues caused consumer dissatisfaction, and warranty / damage claims in many cases. As it grew more regular it not only produced monetary but man-power costs.

To counter the skills gap the company had amassed data analytics programmes and devices that require limited know-how and hired teams of engineers who were familiar with using these. They were not equipped to deal with their maintenance and problems and issues, which required understanding the logic behind the mistakes and fixing them manually within the codes. In time the issues grew in number and depth causing the company to look for expert knowledge. Yet without having to book a selected project with a consulting company or hiring a new employee, both would require substantial time and money. Our expert was introduced to help recapture the value.

Delivery

Entering after 2 years of development, presented an array of projects to fix. It required a mixed approach. Our expert first addressed the critical issues and those escalating, which were pointed out by the manager on an ad hoc basis. These were often caused, because of some bigger programme issues, but quite often started as small bugs, which were too complicated for the staff to fix and they enhanced the problem instead of fixing it. At times these were easy fixes, others complex, some required a new approach, because the issue was not based on the implementation. To truly power the professionals to achieve better results, our expert spent a couple of days with selected data scientists, showing them how to fix the issues and how the smaller interventions work. For projects with more allocated time, our expert acted as an architect by simplifying the paths to meet the modern needs, fixing them from A to Z to root out issues completely and enable expansion.

Such ad hoc case-to-case support was provided by the expert for 3 years, unlocking projects and helping data scientists with troubleshooting their models and deploying them to production. To leverage their potential. Over this period there were also frameworks that had to be crafted and implemented from zero. This end-to-end approach encompassed a number of projects related to the infrastructure improvements as well as those responding to the market situation. The projects implemented during the collaboration period include the anonymization framework created to fill the GDPR requirements, cluster-wide performance audit and tuning, Cloudera cluster upgrades, and supporting dozens of internal Data Scientists and Data engineers.

Effect

All the issues were successfully resolved, and help in navigating their devices enabled the team to realise the value of their solutions.

The post Towering over Telcos. appeared first on TantusData.

The super functional maps.

Magdalena Majka — Thu, 02 Feb 2023 11:40:04 +0000

Client

A worldwide renowned computer and consumer electronics company. One of the largest globally. Headquartered in the USA, it deals with design and manufacturing. Offer includes software and related services and accessories as well as third-party digital content and applications. Both for consumers and professionals.

Opportunity

Operating fast in order to provide relevant updates based on various data sources is crucial for a B2C company building a map service. Achieving efficient updates requires a good communication framework between various teams, among others: Engineers, GIS experts, Data Entry experts. It also calls for compatibility of GIS solutions with big data.

Delivery

Firstly the already implemented solutions at the company were analysed against the review of the working ways of the GIS experts combined with their expectations. They are the enabler of the product, making them the primary interest group.

Secondly, existing data models and their fit together was analysed. It revealed that a data transfer from completely different types of models (object, relational) was the key challenge. The implementation required flexibility to fully support the corner cases inevitably resulting from the complexity of the domain. The delivered solution was capable of transforming every existing data set from and to a relational model.

Effect

The capability to flexibly transform data greatly improved the ability of GIS experts to participate in the work of data engineers. It also dramatically improved the feedback loop of all the changes introduced in any of the existing systems.

The post The super functional maps. appeared first on TantusData.