On my LLM-assisted job search

My love-hate relationship with searching for a job goes back to my first. On one hand, I'm excited by the possibility that lies in the unknown. But on the flip side, I struggle at times with focusing my research exactly because there are so many possibilities.

This time around, LLMs are in the mix. I'm trying to take advantage of them to tap into my excitement, while alleviating my struggles.

Creating a career narrative

I'm somewhere in the middle of my working life, with lots of experience behind me, and more to discover ahead. I've built and maintained a résumé that summarizes my accomplishments and skills, and I can tell plenty of stories about my past. And with the benefit of hindsight, I can identify themes that carry through my whole career, which has had a few interesting twists and turns.

I decided to carry it one step further by creating a long-form narrative. Basically, I wanted to capture all my thoughts about all my past jobs, in as much detail as possible. A kind of narrowly focused memoir. It's something that I could have done at any point, but which is now as quick and easy as it's ever been.

Process

I used ChatGPT for the initial phase. For each job, I just spoke at length, stream-of-consciousness style, dictating as much as I could remember, and relied on the app to transcribe the audio. (I didn't use Advanced Voice Mode, which was cumbersome for the process and unnecessary for the workflow.) Here's my initial prompt lightly edited based on a few refinements I incorporated along the way:

In this chat, we'll be compiling a detailed career history that I'll eventually compile into a set of documents to add to this project. Later, we'll be able to use those documents to brainstorm about career directions, perform targeted job searches, and write custom resumes and cover letters.

Here's the process I'd like to follow.

• I will use voice input to narrate my career experience, one job at a time. As I narrate my experience, keep listening until I've told you I'm done describing that job.
• When I need to take a pause, I'll submit some narrative. Please don't summarize what I've said so far. You can just say "go ahead when you're ready." I'll let you know when I've finished describing that job.
• For each job, you will create a complete description using my narration as raw input. Basically, you'll be cleaning up my language and translating my thoughts into a more coherent form, without removing or summarizing.
• Don't create bullet points; do this in narrative form without veering too far from my language (when in doubt, just use what I say verbatim). Imagine that you're taking dictation for a memoir.
• Don't overelaborate or imagine details that I don't mention. Stick to just the ideas I presented in my narrative.
• We'll compile these narratives into a full description of my career to date.

After this prompt and some initial back-and-forth, I narrated all my jobs. It was pretty darn good at summarizing each one, though I noticed that with some of the lengthier ones, it started to ignore some of my directives, generating bullet points or extrapolating to details that weren't present. I wonder whether starting a fresh chat for each job might have reduced this tendency by increasing the recency of the instructions in the context window.

Output

I compiled all the output into a single Markdown document on my computer and edited it for accuracy. Overall, it was a pretty impressive summary—one that would have taken me probably an order of magnitude longer to write out longhand.

Now that I have this long-form narrative in a local Git repo, I can refer back to it anytime, or fill in extra details that might have escaped me at the time. And of course, I'll keep it updated as I progress to my next job.

I also copied and pasted my chat transcript into another file in the same repo, so I can refer back to the unedited source material. (Unfortunately, ChatGPT makes this cumbersome; I couldn't figure out how to download a full transcript, so I had to copy and paste each individual input.)

Using the narrative

What's the point of having a document like this? I've envisioned a few uses:

Custom résumés. When I apply to a job, I use this document as a source, asking an LLM to generate a résumé tailored to the job description. I'm still tweaking this approach, but the results are promising so far.

Custom cover letters. The same goes for cover letters; the results are also good but need more work to get to where I want them.

I'm sure I'll find additional uses. For example, here's what Claude 3.7 Sonnet says when I upload the narrative file with this prompt:

Using the uploaded career narrative, summarize me in one paragraph: my passions, skills, and experiences, as well as the potential I'll carry into the next chapter of my career. Keep your response to five sentences maximum. Don't cite details or generate complex sentences; keep your summary very high-level.

You are a passionate educator and technologist who has seamlessly blended software engineering with teaching throughout your career. Starting as a developer at Lucent and moving through roles at Peace Corps, Lovett School, Big Nerd Ranch, CNN, and Apple, you've consistently sought opportunities to make technology accessible and educational. Your expertise spans iOS development, curriculum design, and instructional innovation, with a particular talent for breaking complex concepts into engaging, approachable learning experiences. You thrive at the intersection of creativity and technical expertise, whether developing educational apps like tapStory or transforming Apple's coding curriculum. Your unique combination of technical depth, educational insight, and entrepreneurial spirit positions you perfectly for roles that bridge technology development with impactful learning experiences.

Here's ChatGPT o1 says—a different flavor, but also insightful.

You thrive at the intersection of technology, education, and innovation. Throughout your career, you’ve developed software, instructed others, and designed engaging learning experiences. You excel at delivering technical solutions and building accessible resources. Your passion for creativity, collaboration, and continuous improvement drives you to keep learning and growing. With this background, you are poised to make a powerful impact in your next role.

Neither of these says anything I wouldn't already write in a good self-description, but they definitely capture the essence well.

Here's the output from a follow-up prompt:

Tell me something I might not know about myself.

Claude

Based on your career narrative, one thing you might not fully recognize about yourself is that you're a natural systems architect - not just of code, but of learning itself. Throughout your career, you've repeatedly designed holistic systems that transform how people engage with technology and knowledge. From creating a differentiated learning system with randomized mini-assessments in Guinea to developing the structural framework for Apple's web-based SwiftUI curriculum, you consistently approach problems by reimagining the entire system rather than making incremental improvements. This systems-level thinking appears to be an intrinsic part of how you approach challenges, crossing the boundaries between your technical work and educational endeavors, and might represent one of your most distinctive and transferable strengths.

ChatGPT

You consistently turn frustration into fuel for creating new ways to learn and grow. In each role, you look beyond ordinary solutions, shaping innovative paths that others might overlook. This pattern shows a hidden knack for transforming constraints into creative breakthroughs, often before you even realize you’re doing it.

Pretty cool. I'm not sure either qualifies as a genuine breakthrough discovery, but they're both spot-on and definitely thought-provoking.

To be continued

I'm sure this process will evolve along with my skills for prompting and LLM use in general. It's already paying dividends, and it's been fun to do.