Job Search Retro: AI Engineer

Ben landed an AI Engineer Lead role at a growth stage Y Combinator company.

Hedgy Member Background

• Former software engineer who completed a Masters in Computer Science.
• Was focused on AI/ML research or engineer roles.
• Received multiple offers from companies he met through Hedgy.

Key Insights

Running an Efficient Search Process

• If you’re looking for something ASAP, shift into interview mode, put the word out to your network (and Hedgy!) and try to batch a lot of intro/referral conversations together. Getting many concurrent interviews going at once versus a rolling process will help with offer negotiation and interview performance. Once Ben heard he was getting an offer, the remaining interviews that week felt low pressure and he easily got another.
• If possible, avoid starting a search between October-January. Between holidays and annual planning, it’s more difficult to get momentum if you start your search during this period.

Hedgy Profile, LinkedIn Profile, Resume: Don’t Wait to Polish Them

• Hiring managers are doing quick skims of your Hedgy profile, LinkedIn and/or resume at some point in the process. Update each with more results and metrics of my previous work to highlight impact.
• If you’re later in your career, particularly for AI/ML engineering roles don’t feel constrained to one page for your resume/CV.

Spin up a Quick Portfolio

• If you have the time, put together a quick portfolio website so it can show your work and progress in more detail.
• This also doubles as a bit of an interview prep exercise to get ready for the more qualitative or behavioral questions.

What to Expect in AI Engineering Interviews

• Coding questions (often via LeetCode-style problems)
• Systems design (high ambiguity; “show the boxes” kind of questions)
• ML-specific:
  • Debugging or extending PyTorch code in a Colab notebook.
    • It is good to brush up on transformer architecture, especially the concept of multi-head attention. You should be expected to implement this type of component from scratch.
  • Prompt engineering using tools like OpenAI's prompt playground.
  • Writing evaluation functions that score LLM responses—sometimes even using an LLM to judge another LLM.
  • Reading a research paper live for 15 minutes, then talking through it.
    • A good way to prep is to stay up to date on latest models and developments in the LLM space.

This post is part of our job search retro series where Hedgy members who have recently started a new job share learnings from their search. If we can be helpful as you navigate your search, email us at founders@hedgy.works