From February to August 2019, I was able to fulfill a dream of mine: moving to Silicon Valley to work in tech. I had the opportunity to do a six-month internship at Robert Bosch LLC in Sunnyvale, California.
I was working in a small team that was establishing an existing sensor technology on the Californian market. Bosch had successfully developed a leaf sensor which collects data from tree leaves. When processed with sophisticated algorithms, that data allows to estimate the stress level of the tree.
Since the technology had been mainly used for olives thus far, the goal for 2019 was to run a proof of concept study to test the potential of the sensor when used with crops typically found in California’s orchards.
The sensors broadcast data using BLE in set intervals. Central gateways in the field receive those messages, collect them and submit them to the backend periodically. Being the only engineer in the team who actively developed the required software, my responsibility during the internship was to design and develop the entire backend from scratch, to drastically improve an existing gateway prototype based on the Raspberry Pi platform, and to maintain uptime of the system during the data collection phase.
Since cloud computing promises accelerated implementation of complex and flexible backends, we decided to use AWS for our project. Over the course of two months, I gradually developed data pipelines for three different gateway types, a central data lake (S3), a PostgreSQL database that contained processed data (RDS), and a Grafana frontend (Docker/ECS).
After finishing major parts just in time for the phase of data collection, we drove out to various orchards of partners across California. Overall, our 3-person team installed dozens of gateways and hundreds of sensors. In the following weeks, it was part of my job to improve gateway and sensor connectivity, further enhance backend design and implementation, and to build a customer-facing frontend using Grafana.
My internship ended with designing and implementing the data extraction API for the data scientists, who developed the algorithms to extract useful information from the sensor data.