Job Hunting for PhD Students in 2025 – Insights and Reflections from Two Graduates of 'Less Popular' Disciplines

The tech job market in 2025 is full of challenges, especially for PhD graduates specializing in niche fields. In a recent episode of "Li Ding Chatroom," two recent PhD graduates—Tracy and Siqi—shared their personal experiences and valuable insights in navigating the current tough market. Tracy is now an AI scientist at GE HealthCare with a PhD in Human-Computer Interaction (HCI) from the University of Maryland, while Siqi is about to join TikTok as a research engineer after earning her PhD in Computer Graphics from New York University. Their stories offer real-life references for PhD students who are currently or will soon be stepping into the job market.

This blog post is adapted from a recording with Siqi and Tracy. To listen to the original episode, please visit https://lidingzeyu.com/podcast/episodes/job-search.

The Challenge of a "Less Popular" Discipline: When the Market Only Chases Big Models

Both Tracy and Siqi found that, although their research fields are academically deep, in today’s job market dominated by large language models (LLMs), their specialties are considered niche. Tracy admitted that during her PhD, she completed three successful internships (including positions at Adobe and Bosch) and believed that HCI was a hot field—until she started looking for a full-time job and realized that positions specifically for HCI were extremely scarce.

She even suggested an interesting test to determine if your PhD field is niche: "You can check Amazon's Applied Scientist job listings. Since Amazon has such a broad product line, if you find that your resume doesn't even meet their basic criteria, then your field is likely considered niche."

Siqi’s research focuses on traditional computer graphics, such as geometric computing and physical simulation, and she faced the same predicament. She found that positions highly matching her background were "extremely, extremely, extremely rare." Their mutual takeaway was that unless your research directly relates to the current hottest foundation models, landing a job in 2025 will be exceptionally difficult.

Job Search Timeline and Strategies: Casting a Wide Net vs. Precision Targeting

The two guests employed different strategies in their job hunts. Tracy started her search at the end of 2023, after a stint in entrepreneurship. Once her startup efforts concluded, she went full force into job searching after last Thanksgiving, a process that continued until April of this year—just a month before her graduation.

In contrast, Siqi began her job hunt at the start of 2025, planning to graduate in the summer. Her strategy was more targeted. Having received early offers from a few startups and a pharmaceutical company that valued her background in geometry, she was more selective and did not resort to mass applications.

This sharply contrasts with Tracy’s experience. After her initial batch of applications to her top-choice companies all fell through, she became anxious and eventually adopted a wide-net approach, applying to over 200 companies. Siqi estimates that she applied to only thirty to forty positions in total.

The Interview Grind: From Algorithm Practice to Virtual On-Sites

The preparation for interviews was the most energy-draining part of the job hunt, with both describing the process as "extremely difficult" and "exhausting." Tracy estimated that she spent four to five hundred hours preparing—covering algorithm problems, machine learning, system design, and even reading popular papers on large models. This intensive preparation inevitably affected her research progress during her final year of the PhD.

Siqi experienced similar immense pressure, describing how, from early in the year until her thesis defense, she was in a constant state of high alert, even suffering extended bouts of insomnia. She noted that each time before an interview, she needed one or two days to “warm up” on algorithm problems.

Both confirmed that, with very few exceptions for research roles (which are virtually non-existent), almost every company—be it a tech giant, a small firm, or a startup—incorporated programming tasks during interviews, particularly LeetCode-style algorithm questions.

Additionally, all their final round (On-site) interviews were conducted virtually. Tracy recounted that while virtual interviews are convenient, they can sometimes be unexpectedly disruptive—for instance, once her landlord unexpectedly tested the smoke alarm in the middle of an interview, giving her a scare.

Reflections and Recommendations for Future PhD Job Seekers

Looking back on the entire process, Tracy and Siqi shared their experiences and lessons learned:

• Regarding Internships and Return Offers: They both agreed that full-time positions are far less plentiful than internships. Securing an internship in the final year of a PhD is crucial, and prioritizing companies with clear return offer policies is vital. Tracy deeply regretted not opting for a machine learning engineering internship that was more likely to offer a return offer—this might have made her subsequent job search less stressful.

• The Importance of Referrals: Both stressed the critical role of internal referrals. Siqi mentioned that her cold applications received almost no response, while referrals from internal contacts almost always led straight to an interview invitation. Tracy's data showed that the success rate for getting interviews through referrals was about 50%, compared to only 6-7% for cold applications. They advised job seekers to actively reach out to insiders at target companies through platforms like LinkedIn, as even polite and concise introductions can significantly increase the chances of receiving a reply and help.

• Startups vs. Big Corporations: Their experiences diverged when choosing between startups and big companies. Tracy encountered challenges during interviews with startups; interviewers often questioned the technical contributions expected from an HCI PhD. Meanwhile, Siqi found that startups in the healthcare or robotics sectors were more receptive, given that her background aligned well with the needs of those industries. However, both noted that startup hiring processes tend to be faster, with companies often hoping to onboard candidates within a month.

If They Could Do It Over Again...

If given the chance to start over, Tracy said she would have begun conceptualizing a startup project in her first year of the PhD, as HCI research inherently overlaps with prototyping a startup product. Siqi, on the other hand, would have considered a project more closely integrated with AI in her final year to widen her job prospects.

Despite a process full of twists and turns, both Tracy and Siqi eventually secured positions that aligned with their expertise and interests. Their stories genuinely reflect the challenges and opportunities in the current job market. As Siqi summed up, she hopes that all PhD students will "stay true to their passions, never lose sight of their core values, and not let the external environment dictate their paths."