How do academic researchers deal with the tech employment landscape?

 How do academic researchers deal with the tech employment landscape?

Job searching might take a lot of time because every university and organisation is seeking for something different in the applicant.

A apparently unimportant job posting for a research position at NVIDIA went viral on microblogging site X on July 25. In addition to being familiar with machine learning frameworks like PyTorch and Tensorflow, the role required candidates to have published at least eight research papers in prestigious conferences like the International Meeting on Machine Learning (ICML) and computing vision and Pattern Recognition (CVPR).

Several academic scholars were perplexed as to how the graphic card manufacturer arrived at that specific figure. NVIDIA was finally compelled to apologise, calling the bullet point a "error," and stressing that they did not want to compromise quality for quantity due to the widespread criticism from the community. The episode is only one illustration of how difficult it may be for university scholars to get employment. 

volume over quality

Ankit Rai, an AI specialist employed at an advanced computing lab, stated that "research labs typically have high standards for hiring candidates with research capabilities." "Publications demonstrate that the applicant has a solid foundation and the ability to undertake superior research.

These high publication standards, meanwhile, can neglect other skill sets like application and teamwork and place inflated expectations on authors.

"The NVIDIA requirement of eight journals is obviously silly, specifically for a non-academic job," remarked Arvind Narayanan, professor of computer science at Princeton. Paper publishing and practical abilities have very little in common. 

It might be challenging to comprehend the red tape involved in having a paper published at a conference that is well known outside of the academic community.

According to Abhinav Upadhyay, Technology Leader at Accenture Labs, "submitting the paper with an important conference is a lengthy, exhausting process - most deep learning meetings like CVPR have around 10,000 submissions."  "A researcher spends between 10 and 12 months working on a paper, and it takes three months to evaluate one. Thus, it is incorrect to see anything as having a minimum need of a certain quantity.

network preference

A contentious discussion about whether or not researchers needed to keep personal websites occurred on social media around the same time as the NVIDIA job posting controversy. There are valid reasons to create a personal website, according to some researchers, as a LinkedIn page is insufficient to highlight research output.

Websites are a great concept, but they shouldn't be necessary. In comparison to merely using a LinkedIn page, you can manage your full profile and body of work considerably better on a website, according to Upadhyay.

This would depend on whether one wanted to apply to a tech business or an academic institution, according to Rai.

The application processes for academic roles are significantly different from those for industrial ones, and having a website where you can maintain a portfolio will help your application stand out. Industrial jobs are heavily promoted on professional networking sites. 

However, professional networking sites are also indicating a rise in the number of academic job advertisements over the previous few years," he noted. 

simplifying the path?

Job searching might take a lot of time because every university and firm has a different set of requirements for applicants.

Some academic employment markets in the United States, like the one for economics, are simplified. Both advantages and disadvantages exist, as Narayanan pointed out. The fact that more streamlined also implies more inflexible is a disadvantage. There is less room for format experimentation or for defining the discipline's boundaries with flexibility. Since businesses and employment types differ so greatly from one another, a simplified procedure is generally not going to be effective at all for non-academic positions. 

Upadhyay believes that given how nebulous the research process is, simplification is not an option.

"It depends on the project you've been employed for. For instance, in India, the stipend may occasionally be project-based or government-funded. There are plenty of variations. Therefore, defining a particular set of rules is quite difficult, he said. 

To assess a researcher's merit, a deeper knowledge is also necessary.  It's a really formal and structured method that isn't best for this profession, according to Upadhyay. 

"There are scholars who have taken many courses, but when you chat with them, this doesn't show. He explained the contradiction in the employment market by saying that the procedure had turned into a marketing gimmick. 

"I've also seen PhD students who study with BTech or MTech students get assistance with their coding work," Upadhyay remarked. This eventually hurts the researcher since, once employed, you are expected to contribute significantly to deep learning projects rather than merely serve as a consultant.

A shift is coming

But despite the disputes, the majority find comfort in the idea that things have changed for the better. 

can experience both sides of the equation and have more opportunities." Things are much better now than they were five to 10 years ago. 

I can recall when there were four or five PhD students at IIT Kanpur. There are now a lot more PhD students, according to Upadhyay. 

And the trend of demand is just going up

"In general, PhD researchers' employment prospects in the IT industry are favourable. "Just look at the healthcare industry to see how much more interdisciplinary research is needed, where researchers must work together across multiple domains," emphasised Rai. 

Upadhyay claims that IT corporations have the resources and a skills shortage. Companies want employees who are familiar with AI's fundamentals because the field is under the limelight, he claimed.

Apart from their personal advisers, PhD students have various options for career assistance, according to Narayanan. "Some conferences are also starting to have events like phd colloquia which are opportunities for networking for doctoral students to get guidance and feedback from senior scholars as well as peers," the speaker said.

While the process of looking for a job after receiving a PhD might vary based on a number of circumstances, including the field of study, the need for researchers in a certain field, or the candidate's location, Rai highlighted that developing a personal network and going to conferences can help applicants make better educated selections.