Many workplaces have had to adapt in the wake of the global pandemic, the most notable change being the transition to a remote first workplace. More than being an emergency measure, though, this experience has served as a test run proving to many companies that remote work is not only feasible and many of the fears employers used to harbour unfounded, it is a downright boon to productivity and a employee happiness.
Deep Learning companies like most tech companies were ahead of the curve when it comes to remote work, however it still makes a difference whether you are applying to a startup or tech company or to a role in a more traditional industry or academia. In healthcare or banking compliance and safety regulation often make it difficult to have an entirely remote setup, if the company has to ensure that physical access is secured according to some safety standard, they can't very well allow their clients to retrieve and store said data on their local machines. Often traditional management styles that rely on in person interaction and colocation stand in the way of fully remote working arrangements. We expect that to change, however, as management theories of the day shift more and more from supervision and control to alignment and autonomy.
There are some important downsides to remote work, however, which only emerge after the glow of being able to work in your pyjamas and have lunch with your spouse has faded. Most obvious is the social and professional isolation one might experience. Many employees have reported rarely leaving the house during the week, let alone have serendipitous meetings or interactions with colleagues. After a while that can take a psychological toll.
It also means that that transmission of company culture is no longer automatic and implicit, through setting examples and observation, these things now have to be codified in some form and that can take a lot of levity and make it feel forced and unnatural.
In the context of deep learning another important aspect is the role of in person meetings to brainstorm new ideas, discuss research, white board or simply bounce ideas off of other deep learning professionals. Because so much of what one does in a deep learning centred role is breaking new ground getting input and testing one's ideas with the help of colleagues is invaluable. Simulating this environment digitally is difficult.
To tackle these types of problems some companies, for instance
Grammarly, have devised a hybrid model that includes time for the type of face-to-face collaboration mentioned above. These companies still have office, that take on the character of hubs in different locations where employees can congregate. Regular off-site events provide the opportunity for team building and deepening relationships between employees.
While there remain challenges for remote workplaces employers have realised that in order to stay competitive remote work needs to be part of compensation mix. For employees this means that location was replaced by a degree of freedom that can be used to optimise other things, such as how interesting the problem is or the company's mission.