When Ametller Orígen contacted us, they wanted to build a new IT infrastructure to support the fast growth the business was experiencing. They needed a system that was reliable and scalable, while easy to operate and they did not have the time to grow an in-house team to build it.

The project

What started as designing the architecture and back-end applications for the POS administration system, turned into a large-scale undertaking behind all major business processes of the company (sales tracking, inventory management, HR, management of pricing and product catalogue, tracking of marketing and sales campaigns, etc.)

In addition, because of their fast growth Ametller Origen needed a new platform to build their tools on, that allowed for fast iteration and adaptability to changing requirements while providing rock-solid performance. They did not want to be locked-in to a single technology vendor, thus they also wanted to have their own staff involved in the development of the new platform.

Because this new system was so pervasive, the client knew they needed to develop the new software following Agile principles and wanted it to be built on the JVM. Agilogy was a perfect match because we’ve always worked based on Agile methods, we had previous experience in building software on demanding platforms and also importantly, with co-sourcing. We proposed a micro-services architecture based on RESTful APIs and built them with Scala in a way that their own developers teamed up with Agilogy and learned (and can keep learning) by doing.

Some of the things we built for them:

• A CRM system for the loyalty program that can broadcast sales and other information in real time, gathers historical data and is used in the company’s MIS.
• A PoS backend that also tracks cash transactions to increase security and categorizes all this information by date, product and store before it’s fed into the ERP software.
• A user management service that acts as a centrally managed, secure repository of users containing the information, powers of access and authorizations of all personnel in the company.
• An analytics platform containing all historical information (Big Data) on sales, purchasing, vendors, etc. which is updated almost in real time. This is a powerful tool for Business Intelligence and serves many needs and very specific questions of management.
• Inventory management that can track very specific lots of fresh produce, date of entry/date of sale, waste, returns, etc.
• A mobile hybrid application using pictures which helps area-managers in their daily quality control. The information they supply is then used by purchasing managers in the back-office to make better buying and logistics decisions.

Technology stack

Back in 2012 we knew that we wanted a language that improved developer productivity over java 6/7, and this is why we chose Scala. On the other hand, we did not have as many Scala libraries as we have today, and this is why we chose a well-known and stable technology (JAX-RS) for the backend. We also needed to build a backoffice and chose AngularJS.

This almost-a-decade-old code is still being used and mantained by the client.

Remarks

Our developers work with the client as a single team. The platform has enabled quick iteration that adapts to business speed, even ten years after being developed. The stack and the cloud deployment has been key to keep the operation costs low.

Our client (FCC Medio Ambiente) had been developing an in-house platform for "low-code" application development for more than a decade. They wanted to modernize the server-side code base with as little disruption as possible, and they decided to write a new API in Scala that wrapped the legacy code that would be eventually replaced by Scala code.

They also wanted to modernize their development practices introducing some agile techniques like test-driven development, continuous integration, etc.

The project

The core of the system we developed was a RESTful API for the new web application (an AngularJS Single Page Application), that, in time grew to become a system where non-developers could define their own HTTP services and expose them to third parties to enable data integrations. They also were able to ingest data from a variety of sources without developers involvement.

With this system, our client managed to reduce their time-to-market for new integrations while improving the quality of the system and significantly reduce the number of defects found in production.

A major area where Agilogy brings value to the client’s IT team, lies in our "full-stack" and "multi-stack" expertise, serving as a bridge between developers of Frontend and Backend or between engineers of different areas of the Backend of the system (for example, integration between Scala and Jython).

On this very elaborate project we’ve also made substantial improvements both in terms of streamlining the software development process as well as modernizing certain programming practices. We’re leading the implementation of best practices and better methods that allow for more frequent and reliable deliveries of software, while also offering support to the automation of testing and quality control.

Technology stack

As the old system was based in java servlets and cocoon, we chose Scalatra as the base for our new API. We also developed many internal libraries like an interpolation-based jdbc wrapper (a-la doobie), integration with Jython or a distributed queue system that could work over beanstalkd or redis.

Remarks

Bringing a team from Agilogy not only got the job of creating a new API done, but also helped change the way the whole team worked, enabling the adoption of agile practices in every area (frontend, backend, QA and operations).

Better costumer insight through segmentation & targeting

How do you build the software that helps a client analyze BILLIONS of data points to effectively find useful meaning in their Big Data?

Big businesses collect billions of customer data points including purchases, locations, page views, and phone calls. Most of this data isn't very useful on its own; and most businesses don’t have the ability to easily combine such diverse data, let alone make sense of it in real-time. That's why Intent HQ uses algorithms to build profiles that help clients understand customers as unique individuals, including their interests, personality types, and emotional drivers.

At Agilogy we helped Intent HQ develop the software needed to achieve their goals. We contributed to two key features of their product: the Topic Graph and the Insights Dashboard.

The Topic Graph comprehends a series of artefacts that represent the human knowledge and helps understand the affinities between topics and how people relate and interact with them.

Generating the Topic Graph was a process that involves obtaining and parsing a vast amount of data from different sources, Wikipedia among others. After the collection of the data, it is then analyzed and processed to extract human interests and establish the relations between them. The data is then polished to more accurately reflect a human-like understanding of each end-customer.

The Insights Dashboard is a web application that provides Intent HQ's clients with an easy way to access their customers' data and analytics in real time. This tool can help companies effortlessly answer questions like "What are my female customers who live in New York interested in?" or "How do my lowest and highest value customers compare in terms of age?”

The Insights Dashboard has a very unique but simple interface connected to a powerful backend programmed in Scala that is in charge of all the complex algorithms and computations that have to be performed on the data in order to provide reliable insight and statistics.

In essence, we helped Intent HQ to deliver technology that makes vast amounts of consumer data points understandable, meaningful and actionable to their users of the system. Not to mention more user-friendly and in real-time. This is easier than it sounds when you’re dealing with Big Data.

If you’re supplying petroleum products and fuels to thousands of gas stations nationwide and then to specific corporate clients, you (and your clients) need to know the inventory to the last drop every day. In such an environment, how do you manage to share specific inventory data with your clients in a safe, cost-effective and user-friendly way?

Sometimes the technical solution to a complicated problem doesn’t need to be very complicated itself. This can be the case when you need to collect a lot of data from many different sources but the reported answers you’re looking for, aren’t that complex and in fact quite limited. But you will need to think about, design and develop the right tools to tackle the problem.

This is what Agilogy did when Repsol, Spain’s largest petrochemical company, approached us to develop a solution to the following problem: They serve the needs of many public filling stations all across Spain but also some large private clients who have their own proprietary filling stations (think of companies with large fleets of cars or trucks). These clients have their own special needs and requirements on how to read the data.

Amongst other data, it’s important to reliably know the exact inventory of each fuel product in the customer’s tanks nationwide at the end of every day.

It’s not like Repsol didn’t have a reporting system before. The problem was that it was based on dated legacy software that exported standard CSV files. It wasn’t easy for clients or other third parties to look up certain details about remaining inventory, levels of use or volumes.

The reports needed to be generated and extracted in a certain format, copied to a server or sent to each client who in turn needed to import these files and integrate them into their own system. It was a tedious and time-consuming process.

Looking at this problem Agilogy, together with the responsible managers of the client, first designed and then developed a REST API that could be used with their existing metering equipment and sensors to integrate the entire data gathering and reporting process.

Once Agilogy drew up the requirements and specs, we decided to develop the API in Rust with an Actix framework. We used ‘file watchers’ to import the client’s existing CSV files and then re-worked the data to be published in the API we built.

This resulted in neatly streamlined process which made it easier to read and analyze by Repsol and their clients. All they have to do now is connect to the REST API and transfer the data to their own systems directly almost in real-time, with all kinds of detail about the product categories of fuels.

While there are always challenges in any software project, the parameters were so well defined that building the entire application took less than a couple of weeks, including testing.

The result is that with relatively simple, custom built software the client (and its end-clients), are now spending much less time gathering and analyzing data and it takes much less manual labor.

Letgo is a secondhand marketplace app. Its aim is to make useful, quicker and simpler deals between buyers and sellers. By 2018, Letgo was the second fast growing app in the US with more than 75 million downloads (+100 nowadays) and was the most valuated Spanish startup unicorn (around one billion). They have presence in several countries, they are leaders in Turkey and they manage a huge amount of real time data with more than 400 million listing, tens of millions monthly users and 13 million daily chat messages.

In that context they were looking for talented scala engineers who could deal with technical challenges in a fast changing environment.

The project

We were part of their growth team as Scala backend engineers. We helped them to build and maintain a bunch of event driven reactive microservices in charge of campaigns creation, serve push and email notifications, search alerts management and users' categorization.

Together we defined a way to AB test experiments in the backend side.

We strongly contributed to redesign some parts of the system by simplifying them and also by ending a full transition to pure functional programming. By doing so, we were able to deliver faster but also optimized and less error prone services which saved important infrastructure costs.

Technology stack

We stopped using Akka and Scala futures and we ended up using the full cats ecosystems' toolset. The big picture includes: Cats, Cats effect, Doobie, Fs2 streams and Http4s.

Main data infrastructure technologies used were: MySQL, Elastic search, Redis, Influx DB, Amazon SQS and Dynamo DB.

We embrace DevOps culture, so most things were automated and traced. Since CI/CD processes using sbt and Jenkins, infrastructure deploys with Kubernates and Terraform and logging and monitoring dashboards with Kibana and Grafana.

Remarks

Our engineers were game changers in the team. Strategically, they improved some agile processes and helped to adopt some agile software practices like pair programming. Technically, they pushed the team into some technical achievements that were a reference for other teams. We end up spreading and mentoring some of these concepts across the entire company and building inner software libraries for internal use.

How do you reduce delivery times in a city while also improving the customer experience when you need to deliver 240.000 packages a WEEK?

This is the challenge facing Stuart, a delivery service company in over 70 cities with over 3 million packages delivered in 2018. Actually, Delivery service doesn’t quite cover it. Stuart focuses on the ‘the last mile’, operating for the largest part in densely populated cities like Paris, London, Barcelona or Madrid through independent couriers.

You think ‘same-day delivery’ is fast? Try ‘same hour’. But how do you get such a massive amount of packages delivered quickly while also realizing:

ease of use for the customer
environmentally friendly transportation
less congestion in the cities 

You’ll need some state-of-the-art technology using smart algorithms and to build those, you need some great developers. That’s where Agilogy comes in…

Stuart realized from the start that their business would be built around great software. Technology is key to their business model so like many tech start-ups they got together a great team of developers. They built magnificent proprietary software but as they grew quickly, they started running into staffing problems. The IT challenges were becoming numerous and finding the right talent was taking too much time to stay on their projected growth path. Not being able to move ahead was starting to cost real money.

A big challenge was on the Backend side. The code is built in Scala making use of the Akka toolkit. During this process they also built extensive libraries to further develop their complex algorithm. It’s complicated stuff and they couldn’t find the right people fast enough. Since Agilogy specializes in Scala, it was almost natural that we would meet.

To make a long story short, we provided the IT talent to help them accelerate the work on their algorithms in what they call the ‘dispatcher’, the central engine that assigns the invitations for delivery jobs to the ‘drivers’ in the most efficient way. Those drivers can then choose whether to accept or decline a job based on price, convenience, etc. They are completely free to choose.

The recipient in turn decides when they want the package to be delivered and they can follow their package in real time on the app and know the Estimated Time of Arrival.

This may sound simple, but imagine doing that for thousands of packages, 24 hours a day in real time under different traffic conditions, weather, availability of drivers, different allowed weights and volumes of packages, regulations, etc. etc. Like we said, it’s complicated.

In the near future the goal is to bring Machine Learning into the engine by taking actual experience from drivers in real situations and adapting the routes on the go.

Agilogy is helping Stuart to improve their algorithms and realize their business objectives in an agile way. We have no doubt that better IT talent makes better systems and in the end faster and more efficient deliveries.