Arlo is a home video surveillance system, which is evolving into a complete Smart Home system. Recently, More than 7 million devices have been sold. The project continues to actively develop, the product is under testing at the ITS Partner’s R&D center.
The backend is multimodule: it supports both direct possibility to control devices and many related services and integration with third-party products (Amazon Alexa, Google Assistant, and others).
ITS Partner’s TeamLead Egor Pavlovets about the Arlo project:
“This is a unique home and corporate video surveillance system based on artificial intelligence technologies.”
Egor, what do you do at ITS Partner?
I am a lead test automation developer, the head of the test automation engineers team. I have been working at ITS Partner for almost 4 years. My main domain is test automation in all its forms.
Our company works in close cooperation with the American video surveillance market leader – the Arlo company. During the joint collaboration with Arlo, ITS Partner managed to create a unique product for home security and ensure its quality. My team is responsible for quite a wide range of critical functionality: from load testing of cloud-based artificial intelligence server solutions (Arlo cameras provide the ability to detect events in the video stream) to testing the cameras’ firmware.
The automation team leader works in different technology domains, in different programming languages. This is a difficult role, which requires high qualifications and a lot of dedication to the project, product, and the team. But I am very lucky, all the members of our team are high-level specialists.
Tell us about the Arlo system. How does the system work?
Egor: May I ask you another question: Have you ever searched for anything on the video surveillance system?
Asema: A while back, my bicycle was stolen from the hallway of the apartment house. We contacted the cooperative of apartment owners, there were cameras, but it turned out that they had not recorded the moment of the theft. Just think of that: the house manager said that the memory was full. Note that we paid about $ 1 (400 tenges) monthly.
Egor: How does Arlo win the market? You could not find the moment of the theft, you ran out of storage space, there is paid subscription. Arlo solves this at affordable prices for the corporate and private sectors. A continuous video recording (CVR) function constantly records all the events. With the help of artificial intelligence, Arlo puts tags in this vast video stream of information: here was a man, here a postman was passing by, here a car passed, there was an animal, and there was a loud noise. It is much easier and faster to find what you need on such a timeline with events.
The camera breaks down all events by dates, records what happened, and marks it on the time of day schedule. I have a test camera in my office: I can see that at 12 AM someone was in the office or whether the cleaning service was there. You do not need to playback all 24 hours of records, artificial intelligence finds and deteсts these events from the video stream. Another cool point is that the Arlo cameras can work for more than six months using one battery, and using an enhanced one, for a whole year!
If you have a private house, then you do not need to pull any wires, you simply install the camera on a magnet, and your video surveillance starts working. You can track all events on your mobile phone. If there is a person in the yard, you will receive a notification about this and the system will offer a choice: call the police, turn on the siren or spotlight. In areas where wild animals are common, you can set up an automatic rule: during the night hours, if the camera detects an animal in the backyard, turn on a powerful spotlight. Animals are frightened by bright light, and people are not affected by backyard noise while sleeping.
Are you responsible for product testing?
Yes, we are focused on the product, on its quality, new features, and their early release to the market. We perform development and testing in the field of artificial intelligence, this is the research and development of neural networks, which impact so much the video surveillance market. Our team has a cool specialist who does load testing from inside the cloud. A huge challenge for him was creating a simulator of Arlo cameras to generate tens of thousands of connections. Believe it or not, this simulator and all the load testing of the cloud infrastructure was done by one person, I have much admiration for him: he is a genius!
Just imagine the level of tasks for one engineer: create a Linux version of a simulator of Arlo cameras based on real firmware; generate 40 thousand Arlo users with a dozen cameras in the account; within a week, create video streams for all these farms to the servers of the cloud infrastructure; automatically collect all metrics and server health indicators; at the output, generate a report at the NASA level, not only in terms how it looks but also in terms of content. There are only a few such people on the planet.
Thanks to this tremendous and complex work, we are able to perform confirmation testing and provide the Arlo management with answers to whether they are within the planned and promised to consumers parameters of speed and accuracy of object detection in video streams.
How does the R&D center work?
The work of the R&D center can be divided, as expected, into two large areas: research and development. Those research results that have passed and received the approval of the Arlo management go into development. Actually, the development center implements what has been approved as a result of research. I can give a vivid example from our recent R&D project in the testing department: one of the new products successfully passed all the tests, but there was no full confidence in it. This device was handed over to us for testing. To do this, we have assembled a fairly large test bench, where now, using Arduino and a complexly integrated infrastructure test, it is possible to generate various external events: the user has rung the bell, a movement occurred in the visibility area of the passive infrared sensor.
As a result of several months of tight tests and collaboration with all the teams involved, we discovered serious faults in the design of microcontrollers inside the device (the device did not keep the declared characteristics, overheated, and excessively consumed battery power). As a result of this research and retesting, after changing three hardware versions, in the end, all tests turned green and confirmed the compliance of the device with the declared characteristics, and it successfully went on sale. In the development, the process is about the same: first, a concept (PoC – Proof of Concept) is developed, then after a series of coordination and approvals, it goes into industrial development.
The Arlo project is divided into many domains: mobile client for iOS, for Android, the web client, the server side, cloud infrastructures, network issues with video streaming protocols, and testing. In each such domain, we have a team, research, and development.
Egor, tell us about the project team.
A unique group has gathered here, it happened so that in this team, there are people who 10 years ago together with me were junior specialists in other companies. Meanwhile, we worked in fairly large enterprise projects. With these guys, we made up ITS’s expert core and the backbone of the team. These guys are ready to share what they know and help young people improve their knowledge. We decided to rely on junior specialists whom we can train from scratch. And in three years our team of three people has grown up to twenty-one.
Five people in the team are planet-scale experts, they are specialists with 10 years of experience who have gained insight into this domain and can transfer their knowledge further. Four team members have already grown up to be excellent senior managers, five of them are growing up to middles. The rest of the junior + team members are people who have been working for the first year but are already capable of performing independent tasks.
20 % of my team are girls. Just three years ago, on my Linkedin page, I raised the following question: “Do you have at least one female test automation engineer in your team?” Three years ago it was almost impossible to find a female test automation engineer, now the picture has changed radically: there are many female test automation engineers and over these years, the total number of test automation engineers has increased. I see that in some Silicon Valley companies, the demand for test automation engineers exceeds the demand for manual testers.
So you can join ITS as a junior specialist?
Yes, of course. Moreover, we are hiring talented persons with no work experience at all. Because we are an R&D laboratory, the psychological climate is crucial to making a close-knit team. We invest in long-term, serious partnership relations with our partner customers and employees. Therefore, we value personal qualities more than the person’s skills.
Investing in young people is noble work, and I am happy that we can afford it as a company. Thanks to our wise management and all company’s employees, an investment fund has been accumulated. We can pay salaries to trainees without experience so that during the training period, they can fully delve into a new field. Now we are working to make this fund even bigger and free up even more time for mentors to work with the trainees.
As you know, ITS Partner stands out for its high corporate culture, rich values, and long-term mission. We strive to develop and disseminate all modern social trends that have proven their validity and effectiveness. We made such a breakthrough when a new HR director, Evgenia Ivanova, joined the company. She has a prestigious European university degree, and she turned our little cozy world upside down. I am very grateful to her that she was able to explain in simple terms the essence of the values which have long been de-facto the norm in Silicon Valley and which still a little bit frighten us.
We are a bridge between California and Belarus in terms of human values and professional development culture. Now is the time to further build this bridge to Kazakhstan for those engineers who share our values and want to be part of our big job. Now we are opening offices, creating new teams, and not only introducing a professional culture but also disseminating advanced high-tech knowledge, not only taking care of the welfare of each of our employees but also providing them with the opportunity to work on the planet’s top projects.
How to join your team?
Sustain 10 table tennis rallies (smiles).
We have an internship program, which also works in Kazakhstan. If a candidate has successfully passed the test assignments, they join us for an internship. We develop an individual career plan and an individual salary plan for each specialist. The trainees have a complete understanding of where they are now, where they are going, and where they will be in 3 months and in a year, what salary they will have, and what technical skills they need to master to achieve a certain level. We work it all out individually with each of our trainees. Many of them sign three-year contracts with the company, having no intention to leave.
All our internships are paid. At the start, the salary is minimal; this is the time to get to know each other and check if we are suitable for this person. If everything is fine, we close the internship period early and increase the pay five times. After the end of the probation period, the payment is raised again: it is doubled.
Where can you take test assignments?
We have prepared test assignments for your readers who would like to try assessing their strengths.
QA tests: required qualifications, test assignment.
QAA tests: required qualifications, test assignment.
These assignments can be used as an example; usually, we develop new ones each time. Using these tasks, you can test yourself to understand how ready you are for an internship and changing your profession.
What if a candidate fails the test assignments?
I understand how important feedback is. We give feedback to all our applicants who have spent their time on the test assignments. Some of them put a lot of effort, especially beginners. For example, it takes some people 1.5-2 weeks to do test tasks which are normally completed in 1.5 hours. Since we respect their efforts and time, we write feedback for each applicant: what went wrong, where they could have done better, and what they need to read or study. We have templates for some things because there are typical mistakes, but the team members write individual feedback and check all the tasks for some candidates. We know that a persistent and motivated person will return to us for the next recruitment campaign.
Where to study to be a tester?
QA does not sound as cool as a developer; many people do not understand the unique benefits of this profession and its difficulties, they think this is some trifle. Those who have deep knowledge in systems do very well; there are successful cases when guys came after studying at a medical university. I explain it by the fact that those who have learned to take pains during their studies to achieve results and those who like hunting for bugs start very well in the QA profession. Persisting in seeking the root cause is not for everyone. Guys who continue to struggle with the problem when something goes wrong for a week, a month. This profession requires a lot of motivation, try again, make mistakes, and try again.
People who are good at self-study complete different courses successfully. There are many schools; the most important thing is choosing carefully and studying the reviews. I wish you stood firm on the quality assurance front so that no bugs slip through.
Source: bluescreen.kz