The Overengineered Resume with Zola, JSON Resume, Weasyprint, and Nix

I don't know that this is the right way to solve the resume 'problem' - I think LaTeX is a far superior choice, yet the author pretty much dimissed it as a possibility.

For me personally, I found LaTeX to be the perfect solution. I have my resume tex setup so I can set toggles to define what gets output. E.g. applying for a manager position, I might keep it brief and more technical.

The resume is modular and can be updated by updating external txt files and not the LaTeX itself. It looks nice, is always consistent, has nice links, etc.

It's optimized for all the ATS nonsense it inevitably gets run through, it generates a PDF, and I've made it near impossible for recruiters to copy and paste and repurpose it without retyping much of it, and I have a tone of tech tricks in their like invisible text that automated systems might see.

If LaTeX itself is sufficient, I can't imagine needing to add in something like Nix and a webserver or how that would be better in any way.

Automating your resume is a bit of a rite of passage for new software engineers. It just feels like a stupid repetitive task.

I've found that I change job summaries so often, that automating it was a net negative in time spent on the thing. So now I just do the same as the digitally challenged: copy a Word file to "resume (DATE).docx" and change the contents as needed.

My younger self would be surprised, and slightly annoyed, at how often I use the "dumb" solutions for problems.

Mine is also automated, but a lot more lightweight.

I - like OP - use JSON Resume [1]. Besides that I just have a Github Action which creates a PDF and also updates a web-based version [2] hosted in AWS S3. My favourite thing about this setup is that if I want to make a small change I can just log in to Github and commit a quick edit in the repo [3] through Github's web-based UI A new PDF is generated and the web-based version being updates automatically.

[1] https://jsonresume.org/

[2] http://geoff-clayton-cv.s3-website-eu-west-1.amazonaws.com/

[3] https://github.com/jefc1111/cv/tree/master

The linked resume is a poor example of the process.

It's neither visually pleasing nor is it more easily readable than other CVs that I have seen.

It's fine for a CV to be visually ugly as long as its readable, or visually attractive in spite of being less readable. You can't fail at both dimensions.

If you're really happy with a CV that looks and reads like this, save yourself all the effort and make a RTF document instead.

Cool project.

That said, I never send the same resume to everyone. I tailor a resume for each job I apply to. This doesn't mean leaving off positions or lying. It does mean taking the work done in a position and giving it a slant toward the target job.

For example, let's say I created a web app that shows a sales dashboard, with stats visualization, from raw daily sales data.

Job 1 (applying for Front-End) - Created monthly sales dashboard web site using React, MaterialUI, SASS, Node, Riak, and D3. Dashboard provided grid-based summary over individual weeks, months, and years. It also provided configurable line graphs and pie charts for various sales metrics. Data pre-processing done with Sci-Kit. Later added sales prediction using Machine Learning.

Job 2 (applying for Data Analysis) - Analyzed raw daily sales data to determine data cleansing needed, and created tool pipeline using Python NumPy, SciKit, and scikit-learn. Integrated pipeline into Riak data source ingest, and then built sales dashboard web site to visualize the data. Provided ML model for sales prediction built with scikit-learn, with XX parameters. Model achieved YY accuracy with only a ZZ mean error.

Same work in both, but I highlight the tasks most relevant to the target job.

Other thoughts are that Latex is a good way to get a well laid out PDF resume (PDF is not the web, you should have two versions of your resume), and I agree with other commenters - the final product needs more polish if it was actually going to be used to produce a resume to send in (I think it's fine as a proof of concept though).

I wish there was a standard format for resume that was universally used. Every HR website requires you to upload a resume to apply, then tries to extract the various experiences and details automatically, invariably fucks it up completely, and you end up having to spend 20 minutes to correct it manually. Unless they treat this exercise as a form of captcha…

Interesting approach. I am currently looking for jobs and went the 'career coaching' route for my CV. I did a few iterations with my coach until I got my current result (ideally I had a link):

I first looked at Canva templates, but apparently nowadays you are supposed to do black/white and no fancy designs for ATS readability. Then I tried it with Google Docs b/w resumee template, which kinda got me to write actual skills. Then I approached the coach, got her template and iterated, and then I also added some rules from here (https://principiae.be/pdfs/ECV-1.01.pdf).

I also involved ChatGPT to analyze job postings and to get the mix of keywords in my resumme right. Tools like https://tagcrowd.com/ also help with that. For example, I am targeting 'IT analyst' roles, and it does make sense that I have the word 'analysis' a few times in my CV.

E: mine is basically structured the following way

Name

Title

Summary

3x5 ATS keywords/skills specific to my profile and role

last ten years, also written in a way that 'gamifies' ATS: 'Year, worked as ROLE at Company, did XYZ'

--page 2--

Education (degree + grades)

Skills Training

Languages

Some more IT skills (programming languages, project management, ...)

E2: I obviously have no idea what I am doing, but I got three interview proposals for 10 applications, so I guess 30%.

As web dev, I migrated from TeX to svelte + Chrome's Print-to-PDF.

I've found that splitting data and representation is not as feasible as it sounds. You add a job and suddenly your CV doesn't fit on a page anymore, and cutting details from previous jobs isn't enough. So you change the layout, ever so slightly, because the data changed. And version control? Nice for building, but it's not like you'll ever go back in time anyway.

What's nice though is defining the data and then trying different layouts to see what works.

I have a resume in latex, all I have to do is change a couple lines every now and then and run the default latex pipeline, all of it directly from gitlab. It worked on the first time I tried and every time since then, it produces a pdf which I can then download and send.

I noticed the container they run uses nix too, which is nice although I don't care about it as long as it works. I could add signing of the pdf maybe some day for fun. What's great with this approach is how little hassle there is and nothing to install, nowhere, and produces the same clean resume I've used for over a decade (but needed to install a thousand things I could never remember from one computer to another).

The only thing you should spend time on with your CV is the content. The ROI on tweaking the look and feel is very low. Aesthetics are not what people care about, unless your resume is so hideous that it gets thrown out.

I had the same idea sometime ago.

Not sure if you can say that it was over engineered or not but I used the following:

* Frontend framework : Next.js / React (Functional components with React Hooks)

* Rendering : Static Site Generation

* Programming language : Typescript

* CI/CD : Github actions

* Unit test : jest

* Design : SASS / Responsive design

* Data validation : AJV / JSON Schema / Joi

* Infrastructure : Cloudflare pages / Terraform

* Package management: Yarn

* Linting & Formatting : ESLint / StyleLint / Prettier

* Pattern matching : ts-pattern

* CSS framework : react-bootstrap

* Monorepo : nx

* PDF generation : jspdf

* Contact form : web3forms

* Captcha : hCaptcha

I am quite happy with the final output : https://www.remikeat.com

It would pull the data from

https://data.remikeat.com/resume.en.json

https://data.remikeat.com/resume.jp.json

https://data.remikeat.com/resume.fr.json

So I can just update the JSON and the webpage will update itself.

Also as the PDF is generated locally, the PDF also get updated automatically.

And I didn't know there was a JSON standard for resume. Maybe, I should migrate the format I designed to this open standard.

Ultimately, I wanted to add a portfolio section, where I would show some of my projects like

https://stackl.remikeat.com which is a stack language interpreter written in Ocaml and compiled to js with js_of_ocaml.

I also over-engineered how I generate my CV[1], but went the opposite direction by using Dhall to create JSON and LaTeX files that I use to create a PDF and GraphQL API in Rust for it, automatically deployed via CI/CD to a VPS and a tagged GitHub release. It was a lot of fun to make, but is so over-engineered I hardly want to touch it anymore :)

[1]: https://github.com/sondr3/cv-aas

I did something similar but with my phone, the android orgzly app, termux, jinja2 and latex.

I cribbed a CV template from overleaf and put some jinja2 in it to take the content from CV note in my note taking app: https://hitchdev.com/orji/using/latex-cv/

With termux I can then hit a button to run a script and it will instantly generate a pretty PDF using latex from an orgzly note and fire off an android share intent.

The nice thing about this set up was that if a recruiter called me while I was out and wanted a CV quickly with a couple of tweaks made I could just do it on my note taking app and email an updated PDF in a few seconds.

In theory I could easily change the style of the CV but in practice I haven't felt the need to touch it in years.

It is definitely overengineered and, unfortunately, outdated, as the most recent position in the resume and about page do not match. A lot of effort for something that is not accurate and obviously not used to apply for jobs, which is the purpose of a resume in most cases.

A simple file (Word doc, Numbers, Google Doc) that receives a change every few years and is exported as a PDF seems to do the task better.

The blog set me up for a really gorgeous typeset resume with lines in the intro like “More visual flair and typesetting control”.

With that expectation set, the end product was severely underwhelming. The visual flare is non-existent.

There's also Manfred's MAC schema for describing resumes https://github.com/getmanfred/mac

The overengineering here produced a poor result. I have found a lot of success using the pandoc_resume project, which is literally just a content item written in Markdown, formatted into different outputs using a LaTeX template via `pandoc`. With this, I output a PDF that looks great, and output HTML which I put in as a non-touched file in my static site generator, and it works very well.

Typst a way simpler approach to this:

https://typst.app/

This reminds me a bit of that animated Flash résumé that some young animator did, back in the day. It received quite a bit of attention (both good and bad).

It was basically a cartoon version of him, walking through his various life accomplishments.

It was well-done, and all, but I found it a bit annoying. Also, it was Flash.

I think the presentation of a resume is much less important than the content. It’s interesting that sourcers and recruiters make most of the decisions about who gets to the interview stage (particularly in a down market), yet they are the least qualified to assess the capabilities of the candidate of everyone in the whole process. Despite that, their assessment of a candidate’s resume has an outsized influence on the candidate’s outcome.

When you have 1,000 resumes from laid off software engineers from FANG companies that all look mostly the same, how do you decide who to call?

I’ve been thinking lately that an interesting project might be to look for publicly available resumes of people who have recently accepted new jobs within the tech industry and compare their resumes against those of people who have been looking for a new job for a while. The comparison would be qualitative if only a few resumes are available or perhaps quantitative (i.e., a classification model) if many are available.

Recruiters are not looking for the same signals as hiring managers, and since I’m not a recruiter, I would really like to know exactly what it is that they are looking for.

As an example of this kind of discrepancy that caught me totally off-guard, I was slightly below the “years of experience” requirement on a particular job posting that seemed to match my background perfectly. The recruiter I was talking to had reached out about another job posting where I did meet the YoE requirement, but I said the other posting was a significantly better fit for my skills and experience. The recruiter replied that the YoE requirement was not negotiable, so I was put onto the interview loop for the much less applicable role because of this arbitrary and narrowly missed line in the sand (and of course I failed that interview, wasting both my time and theirs).

A hiring manager would care less about years in seat and more about capability. That such a YoE requirement might be used as a hard filter when scanning resumes caught me by surprise—I had previously sent out plenty of resumes where I narrowly missed the YoE requirement, and in retrospect, my application was probably discarded immediately while using up the “quota” of how many times I could apply to that company. These sorts of insights from the recruiting world would be great to know in advance from the candidate’s perspective.

I started doing something less over-engineered last year. It uses LaTeX with a CD pipeline and I think it's is a good way to do resumes. Every time I push mine, the one linked on my personal website is updated to the new version.

The diffing and version pinning is nice. Throughout intern application season I usually make updates to my resume and it's nice to be able to easily go back to the version of the resume that I applied with whenever I get a callback from a company.

Mine: https://github.com/nickorlow/resume

Website with link pointing to latest pdf: https://nickorlow.com

I also do something similar with cheat-sheets for classes. Using git for collaboration it is really nice: https://github.com/nickorlow/cs-331-cheatsheet

I do like the concept in the article of separating the data from the presentation.

Scrolled to "The Result" section only to see... nothing? Would've been nice to see the resulting resume

I created my CV with React and Semantic UI since I was not satisfied with the amount of visual precision Word provides. The stack choice was due to personal ease of use, Semantic is probably an overkill.

Recently, I started turning it into a template and added some documentation. The next step is to feed the data (experience, education) through JSON so it's easier for others. It is automated to the point of generating a new preview GIF every time I push to master. I still manually hit CTRL+P to get the PDF though, it feels faster than implementing automation.

I'll hopefully do a Show HN in the future but until then, here is a demo on GitHub Pages: https://mrtysn.github.io/cv/

I put together an HTML+CSS template for authoring one-page documents that will be printed out (or saved as PDF):

https://github.com/jkitching/1pager-printable-html

It can be used for creating a PDF locally, or the HTML file can also be viewed directly in the browser - both should look identical.

When I was looking into different ways of converting HTML into PDF, everything I came across did a pretty bad job at correctly supporting CSS layouts and positioning. So I ended up just using headless Chrome for printing. Perhaps WeasyPrint does a better job with CSS support?

Now that's "resume-driven development".

And hey, overdesigned or not, it's a resume that ended up on the front page of Hacker News, making it all worth it given what a resume is supposed to do.

I love me an overengineered resume. I've been hacking on my academic cv as part of my personal website for years and years, with unholy stuff like live TeX libraries downloaded in the midst of a netlify CI process. Still needs some work on various formatting edge cases, but in case you're curious: https://github.com/paultopia/websiterevision2017

I'm working on hacking the job application process by having LLMs take your base resume and tailor it for each job application. I'd love your feedback https://customizedresumes.com . I've been A/B testing by applying for jobs with only my base resume compared to an AI-customized resume. It's roughly a 2x higher response rate with the customized resume.

I love the idea, but the thing that keeps me from over-engineering my own resume is the number of times recruiters have asked for a copy in Word format, so they can strip out all the identifying features. Lots of large companies mandated this practice to combat bias when screening resumes. This irks me because I'm really just trying to get the Github profile and personal site URLs in my resume in front of technical people.

There is a much simpler approach to making a nice looking resume.

Just use markdown > html > pdf using a custom css style sheet

This is what I use for my resume:

https://github.com/LarsDu/LarryDuMarkdownResume

You don't see many people who went to St. John's college. If the author or any St. John alumni are reading this, I'd love to hear what you thought about the experience.

It would be bring less complexity using https://devenv.sh/ to provide the tooling :-)

Can't we all just agree as an industry to use LinkedIn profiles and move on to more productive things? I know they're horrible but come on already.

Am I missing something? Where it says "The Result", I don't see the result. Am I looking in the wrong place?

Afaik modern resume parsers prefer .docx to .pdf. Some won't even accept a PDF.

OMg really hope you got 10x jobs that workflow.

But the results looks... very bad?

"over engineered resume"

oh boy. wait until you see europass standard.

appetizer: it's a pdf, with embedded xml, with html snipet field values.

Reminds me of :

https://xkcd.com/1205/ and https://xkcd.com/1319/

Unsolicited advice:

If you're going to that trouble, finish the swing. The final product here is BAD.

The person reading the resume doesn't know it's automated so no credit points for that, and it looks like a person didn't pay attention to basic detail. This does you no good, it hurts you.

Example glitches:

The contact row dropping LinkedIn to next line. The ragged bullets hanging to the left of the left margin relative to headlines. The dates not right aligned. The too-large font making it take ALL of two pages for just a decade.

These things do get judged at a glance when someone's looking at lots of PDFs.

Also, just because a resume goes into an ATS doesn't mean the hiring manager isn't looking at the original PDF. Usually the ATS surfaces/sorts applicants but one reads the original PDF or Word doc anyway, because the ATS interpretation is often unreliable on its own.

All that said...

One idea to counter the above advice is either in italics under the contact line or in a clear footer, put a colophon or imprint saying: This PDF was auto-generated from my job history data, blogged about the automation here: https://full.web.link/resume-blog-post

But still, exhibit a higher bar for the end product. Engineers that fall in love with over-engineering without regard for the output and "end user experience" are everywhere, and a problem. Engineers that produce above-the-bar output end users appreciate, while ironically over-engineering, with full tongue in cheek recognition of the over-engineering, are rare.

Dropping a link/endorsement for quarto : https://quarto.org/

It's a static site/document generator that supports bibliographies, jupyter notebooks, and good old fashioned markdown. It works well for complex academic resumes and CV's, as well as blogs and library documentation.

I use a template for my website : https://jjd.io/

github repo : https://github.com/jdonaldson/jjd.io

Also, if anyone has any positive/negative comments on my site, let er rip! I'm still working on it.

Shameless plug: once you are done building your resume, copy and paste the text into CommandJobs[1] to automatically match your experience with job listings and get a filtered list of the best ones for you

[1]: https://github.com/nicobrenner/commandjobs

And here I'm, writing the resume in markdown using Obsidian, slapping some custom CSS until it looks good and hitting Export to PDF.

This page is utterly confusing. Is it about a CV? I don't see what is being resumed.