57 interviews for OpenAI: a PhD student revealed the entire behind-the-scenes reality of finding a job in tech

Getting a job at one of the world’s most coveted tech companies turned out to be much harder than many imagine. Computer science PhD graduate Alice Liu shared her path to an offer from OpenAI and published a detailed account of her job search, which is already being called one of the most useful guides for professionals in IT and artificial intelligence.

During her job search, she went through 57 full interviews at 11 companies, held 46 conversations with recruiters, and another 16 meetings after receiving job offers. The researcher described her experience in detail in the article “Notes on the Industry Job Search,” where she not only explained the hiring process but also compiled dozens of useful materials for interview preparation.

Why the job search turned out to be so difficult

At first glance, Alice was an ideal candidate. She spent six years in graduate school at the University of Washington, conducted research in natural language processing and machine learning, published academic papers, and took part in major research projects.

However, even with such a résumé, she had to go through a real interview marathon.

According to Liu, modern AI companies evaluate candidates across several areas at once. Many applicants, in fact, do well at one type of task and fail at another.

What kinds of interviews AI companies conduct

During her job search, Alice identified several main interview formats.

Algorithmic tasks

These are classic programming problems, similar to tasks from the LeetCode platform. The candidate needs to write working code quickly and explain their solution.

According to Liu, even experienced researchers often underestimate this stage and lose good offers due to insufficient practice.

Machine Learning Coding

This format is becoming more and more popular.

Candidates are asked to implement parts of neural networks or machine learning algorithms right during the interview. For example:

- write an attention mechanism;

- implement logistic regression training;

- build a simple neural network;

- explain how transformers work;

- implement a tokenizer.

In Liu’s view, this is the stage that causes the greatest difficulty for candidates right now.

System Design for ML

These interviews test the ability to design real AI systems.

An applicant may be asked to design:

- a ChatGPT-level chatbot;

- a recommendation system;

- a search engine;

- a content ranking system;

- infrastructure for training large language models.

Here, the evaluation is not about code, but about understanding architecture and the ability to make engineering decisions.

Research Interviews

These are especially important for research roles.

The candidate gives a detailed account of their research projects, publications, and experiments.

According to Liu, many employers pay even more attention to this stage than to publications in the résumé.

Behavioral interviews

Despite the technical focus of the industry, companies actively assess communication skills.

Candidates are asked:

- how they resolved conflicts;

- how they handled setbacks;

- how they worked in a team;

- how they made difficult decisions.

Liu notes that many engineers take such questions lightly and make a big mistake.

The main takeaway: a degree alone is no longer enough

One of the researcher’s most unexpected conclusions is that a strong education is no longer a guarantee of success.

According to her, an academic degree helps you get an interview invitation, but after that employers are interested solely in practical skills.

“No one cares how smart you look on paper. What matters is showing that you can solve real problems,” the author concludes.

How she prepared

Liu spent hundreds of hours preparing.

She focused mainly on:

- solving LeetCode problems;

- reviewing the basics of machine learning;

- practicing ML coding;

- studying the architectures of modern language models;

- preparing stories about her own projects;

- practicing answers to behavioral questions.

She recommends not memorizing answers word for word, but preparing a set of stories from your own experience that can be adapted to different questions.

Why networking turned out to be more important than a résumé

Liu dedicated a separate chapter of her article to industry connections.

According to her, a significant share of opportunities appears not through cold applications to job postings, but thanks to referrals from acquaintances.

She advises:

- attend conferences;

- go to professional meetups;

- stay in touch with former colleagues;

- don’t be afraid to message people from companies you’re interested in;

- talk publicly about your projects.

In the researcher’s view, strong professional connections can shorten the path to an interview several times over.

How not to burn out during a job search

One of the most valuable sections of the article is the advice on dealing with emotional burnout.

After dozens of interviews, even strong candidates begin to lose confidence.

Liu recommends:

- not taking rejection as a judgment of your own worth;

- remembering that many decisions are made based on the company’s internal needs;

- taking breaks between interviews;

- not preparing around the clock;

- leaving time for rest and socializing.

She admits that she herself felt tired and disappointed many times, especially after rejections from companies she really wanted to join.

How to get the best offer

After receiving the first offers, the work is not over.

Liu emphasizes that many candidates are afraid to discuss terms, although negotiations are a normal part of the hiring process.

She advises:

- not to accept the first offer right away;

- compare several offers;

- clarify bonuses and stock options;

- discuss compensation openly and professionally;

- use competing offers as leverage in negotiations.

According to her, smart negotiations can increase total compensation by tens of thousands of dollars per year.

What ultimately helped her get into OpenAI

In Liu’s own view, the decisive factors were not her degree or publications, but a combination of several things:

- deep understanding of machine learning;

- systematic interview preparation;

- the ability to explain complex ideas in simple terms;

- professional connections;

- persistence.

Her story shows just how competitive the job market has become in the age of artificial intelligence. Even a candidate with strong research experience went through 57 interviews before receiving the offer she wanted. Still, Alice’s experience proves that careful preparation and consistent work on your skills can open doors even to the world’s most prestigious tech companies.