WISRAN

Wisran is a revolutionary startup idea, selected by MIT boot camp. It uses data science to measure time variations of farming.

CHALLENGES

Limited time and budget

Provided with all of its complexity, we had to prove our idea, by developing a minimum viable product, within a season, i.e. 3 months as we didn’t have enough budget to record data of whole year farming cycle covering all processes and seasons.

Real-time activity mapping

As different industrial farming vehicles, each having different sizes, speed and movement actions were involved even in a single process, identification of current activity was dependent on many variables. We had to develop a rule engine for analyzing a broad set of variables and determining current activity and its relative stage.

Scalability

Scalability was a real challenge as adding a single node in the system adds a huge volume of data. We had to design a decoupled architecture that keeps all software components, along the data pipeline and data aggregation engine in their optimally efficient state and ensures the analysis, aggregation, and computation on a vast set of data provides a real-time picture.

RESULTS

Wisran is live with an Android and an iOS app. It has now successfully automated 56713 acres of industrial agriculture fields and made its way to Australia by winning a grant from the Australian government.

Solutions

Limited time and budget

Our years-long experience of managing medium and large complex projects helped us in keeping this project on its planned trajectory. We followed Agile processes and divided relevant stories in sprints with complete focus on adding maximum business value in each sprint.

Real-time activity mapping

We used Apache Kafka and Nifi to receive and channelize data streams, being relayed towards our rule engine powered by Apache spark, that analyzed and made aggregations based on complex rules at high speed.

Scalability

The decoupled infrastructure design rescued us. We designed the system in three layers: a data pipeline, a core engine, and a data layer. Each of these layers is a multi-node system, and we can add multiple nodes in each layer according to the system demand, anytime.

Complex Business Domain

Our onsite manager visited the original site and conducted dozens of interviews with clients, business stakeholders and vehicle operators. Various interactive meetings and brainstorming sessions were held. With the well-organized collaborative effort, we were able to convert business requirements to stories and epics.