Buy me a hot chocolate! | About Me
date: Tues 2. May 16:48:32 CEST 2023
If you find this useful and especially if this post helps you land a job, send me a tip!
May 2024 update:
It is the year of the LLM. ChatGPT, Gemini, Claude, PerplexityAI, etc., etc.
A lot of what I wrote below is still useful … but the hard part, the tedious parts can not be expedited with these large-language-models or LLMs.
So here’s the gist / summary of what it takes to write a good resume in 2024:
Write your worst, wordiest, most verbose resume you can. Include all the details.
Then ask an LLM (I am quite partial to Google’s Gemini but use whichever one – and the free one’s are enough – no need to pay) to make it “more concise and in the language of the ideal resume for role as a” and then insert the role you’re trying to get.
Iterate on this a few times asking the LLM to tweak things you think could use improvement. And then incorporate that into your resume.
That’s it.
I still think there’s value in reading the parts from below – but the machinery to get a really good resume has just gotten a lot easier with these new fangled LLMs.
What a time to be alive!
First things first, I am not, nor was I ever, a professional writer of, or reviewer of, resumes. That being said, I know how the system works. I have been on both sides of the hiring desk as a job-seeker and an interviewer. I know how to structure a resume such that it will get screened well by the automated systems as well as human reviewers.
And here’s the kicker: as a full-time, in-the-trenches data engineer I can see through the filler. While I might get fooled by a really slick resume it’s very hard to talk one’s way around a poor performance on a job-applicable, daily-style task, in a technical interview.
So whether your goal is to impress your next interviewer with an amazing resume or just open some more doors, take a minute and read this post, and see if it doesn’t improve your resume and open any of those doors.
All I offer is my take on what I think the ideal data engineering resume would look like, how it would be structured; how it might flow. It’s not a science it’s much more of an art but I will convey all the morsels of info that I can think of to help you improve your resume and be successful.
The aim of this article is to prepare you to submit a really solid resume that will get you in the door and past the automated bots that will be keyword scanning your resume. Once you get past the door I have other tips but those’ll have to wait for different post.
Follow as much of the suggestions as you can. You might not have days to rewrite and then rewrite again your resume. But this excercise of reviewing what you’ve done, what you’ve learned, and what value you provide to a business is an important excercise to embark on not just for having a compelling resume but also will empower you for when you go into performance reviews, ask for a raise, or a promotion.
I have to tailor this article for mid- to senior level engineers as driving this article will be me refactoring my own resume.
I will however write another article in follow-up to this for those just leaving unversity/college at a later time.
You might be wondering why am I yelling at you? I am not. Sorry. It’s just I wish I had done this sooner. The reason for not listing your phone number is that your future employer isn’t going to call you. They’ll email you. And only as a backup to a video interview would you provide your phone number as a backup.
The company’s HR department isn’t going to refer to your resume for your phone number if/when they lose it, it’ll be in some system and it’ll have gotten there because you filled it in on some form.
By not listing your resume (also remove or hide it from your LinkedIn profile) you prevent folks that would harvest phone numbers. These people could be scammers, identity thieves, and recruiters.
I am not saying that the latter is the same as the former two, not at all.
It’s about keeping control over who has your phone number.
Let the company or recruiter work for your information and show that they’re interested, don’t add your phone number to your resume.
Otherwise your listed phone-number will be scraped by any number of tools and added to banks of call lists.
Again, I am not yelling at you. This is just so very important. In the same vein as not adding your phone number at no point is your employer going to mail you snail-mail things related to your employment from the address listed on your resume.
Would you list your home address for the public to see on Facebook? On Twitter? Then why publically put it out on places like LinkedIn or on your resume?
That’s table stakes now and it’s not worth mentioning. It’s assumed that you know how to navigate a document authoring tool like Microsoft Word and a spreadsheet like Microsoft Excel and be able to author even a basic presentation in something like Microsoft Powerpoint.
Unless you’re doing desktop publishing for a fancy magazine which is very likely done in Adobe InDesign or something don’t list proficency in Microsoft Office as a skill. Besides most places use Google Docs anyway. (Don’t list the Google suite as a skill, either.)
I like reading these. I find them to be the synthesis of larger cover letters. I find that they act as a summary of the whole resume, a kind of abstract, that frames how I will think about reading the resume.
“Create a tailored resume for each role.”
Note on crafting a custom resume for each role: I think that writing a custom-tailored resume for a given role is an effective excercise to do if you have the time. If not, a well worded mission statement and a clear resume will get you far!
You might be tempted to write a cover letter for each role that you apply for. That, too, is very time consuming and unless specifically requested offers very little pay-off. Also cover letter’s aren’t usually requested, read, or even required at most places. Think about it like this:
You’re a hiring manager and have 1 role and 100 applicants. Ideally you would spend copious amounts of time to get to know all the candidates and choose the best ones to go onto interviews. But What you really do is scan all the resumes looking for keywords, key phrases, and maybe spend 30 seconds to a minute on each one.
So spend the most time on your resume as they will surely want one. You can get the best of both worlds by summarizing a would-be cover letter in 3-5 sentences at the top of your resume following the template below:
That last bit about “the org’s mission/values matching your own” is a worth adding if you’re joining a startup with a lofty goal:
Note: If you’re working for a startup or established company doing any of the above let’s talk, I’d love to profile you and the company and what is being done in those areas!
Especially if the company is a startup you’re going to basically be wearing a lot of hats as they say and if you’re technically capable what they will likely look for is someone who believes in the vision because startups don’t always pay well – a lot of the value is on the backend in stocks/options/etc that will grow in value if the idea takes off and the company continues to grow or exits via an aquisition or goes public via an IPO.
Update: Sun 14. May 12:24:57 CEST 2023
Matt Brady: founder at Zuma a DataScience and Analytics recruitment agency mentions that:
IMO; [A mission statement] is powerful for Juniors as the company are hiring [them] for potential. When it comes to Seniors they are hiring [them] for [their] expertise.I think I agree. I think I included it as I still feel quite junior while working on senior level things. I think for my own resume I’ll keep it but I will also keep an eye on it as I continue to retrieve feedback on this bit. I do plan on surveying hiring managers to get their takes on all things resumes. I’ll update this document with that info.
I was told when I was looking for roles in Germany that some hiring managers want to see your face on there. Some folks add headshots to their resumes, even professionally taken ones.
While I appreciate a handsome photo as much as the next person I think the practice of including them on resume’s perpetuates biases and expectations that not everyone can meet. It also stokes reviewer bias. Again, this is not the fault of you the resume drafter. It’s very much an industry issue: there are far too many engineers that look alike on various axes be it gender, race, etc.
Collectively we can help change the industry a bit by removing our photos from our resumes.
My old resume was a bit crap. It failed on a lot of the points I listed.
Before I begin, updating one’s resume regularly (say every 3 months or so) is a good habit to get into. Why? Because what if you go a whole quarter and can’t think of anything meaningful you’ve done? What if you can’t think of things you’re proud of, that you’d be excited to share in an interview? What if you’ve gone 3-months and haven’t learned anything new? That then would be time to reconsider what you’re doing and logging achievements. These could come in handy when going in for performance reviews, as well.
The template I will follow is this:
Name (Job Title) | Email | Website | LinkedIn | Location
Mission Statement
Past Experience
Education
Tech/Skills Acronym Bingo
Certifications
References
Few of us can place our resume’s on a small business card. Titans of industry could. For example: Kelsey Hightower – a stalwart force in the Kubernetes and Golang spaces – doesn’t need a resume. He could just put his name on a card and submit that with links to the talks he’s given, his github profile, and that’s it. The man is a legend. (Okay, okay, enough fanboying from me)
For the rest of us if you’ve been in the industry long enough you’ve probably done more than 1 page’s worth of worthwhile work. Don’t be afraid to list them and exceed a page, just be judicious about the points added.
Here I’ll begin working on my resume and commenting on what I am thinking at each step.
Alex Narayan (he/him) (Senior Data Engineer) | email: [email protected] | website: gigatexal.blog | LinkedIn | Location: Berlin, Germany
I like the above. It’s clear. It lists all the relevant parts. If the reviewer wants they can look me up on the web by checking out my blog or immediately figure out who I am connected to via the linkedin link and is aware of where I am located which should set aside any visa questions.
Mission Statement/Summary: I am a pragmatic and capable engineer with over 10 years of experience in data and data engineering. During my time in industry I have worn many hats: Database Adminstrator, Python Developer, Site Reliability Engineer, and now Data Engineer and have learned from many amazing engineers and mentors. All of my experience from my previous roles has shaped me into a well-rounded, capable, engineer who seeks solutions that are managable, cost-effective, and practical. I also enjoy bridging the divide that often exists between business and engineering teams by being able to communicate simply between the two. I am seeking to join a team where my experience and skills can be of value and also from whom I can learn and grow my own talents.
Aside here, the above is really hard. I don’t even think it’s that good. But I don’t think it’s terrible. I didn’t want to get into the trap of using too many empty business-speak words. I’ll be sure to revisit this in the future. Notice how I reference the previous roles I’ve had? This serves as making the mission statement a sort of thesis for the resume and helps show how versatile I am as an engineer.
What I wanted to convey though is the summary or thesis of my resume and how I think: I am too jaded to be fooled by the new shiny, I stick to tried-and-true, practical tech choices to solve buisness problems because I have “seen things” (I did kinda like that phrasing a lot).
I once joined a startup trying to reshape logistics and we ended up building on the HackerNews stack: Micro-Services, Kubernetes, Cloud.
The solution we should have gone with was a simple monolithic application/api/website in Python (Flask/Django didn’t matter) to prove out the idea. Suffice to say we wasted months with all the complexity. The startup pivoted the development team to India and I left. So what I am saying is that I won’t bring that silly hubris with me I will bring things that I have seen working and have actually shipped.
Let’s continue through the process to figure out what my first set of bullet points will be for my current role. This process I will repeat for all my previous roles.
This method is commonly recommended for when one is interviewing. But I find it useful when crafting your resume. This is my secret sauce as I think it adds color to the resume and makes them less dry but it also tells a story much like one would do in an interview.
Experience:
Job title and start date and end date (“current” if current role): Senior Data Engineer Delivery Hero GmbH June 2020 - Current
Ad tracking - In house tracking - Lots of data, schemaless, curation - High level design - Code Review / Initial design
Situation: Take over stewardship of in-house front-end tracking service
Task: Define the primary key for tracking events
Action: Worked with advertising teams running campaigns, app and web sdk teams, and backend teams to define an optimal bigquery schema and primary key columns
Result: Able to achieve parity with Google Analytics with our in-house service to drive insights and revenues from advertising campaigns
All Together: Helped to operationalize an in-house replacement for Google Analytics by coordinating with advertising teams, front-end and app teams, and back-end teams to define assess the uniqueness of columns (cardinality) via a Request For Comment (RFC) approach as well as ad-hoc investigations and created an optimal schema for incremental batched ingestion of app/web tracking events in compliance with GDPR
The final product of the “STAR” approach is to more or less start backwards: Start with the result and work your way backwards to the problem and such. Notice how I start with “operationalize an in-house replacement for …” and then explain how and the why?
What I like about the above is that it’s not hyperbole: I very much did do those things albeit as part of a team. I was not the sole project leader of this effort but I was part of a team of very highly competent engineers working to tame the beast that is front-end tracking.
“… via the Request For Comment (RFC)” is a sly way of saying: “Yep! I know exactly how large companies coordinate changes that affect many teams and I can write and convey my ideas in a structured way and get buy-off on changes for my designs and improvements.”
What it would look like on the resume:
Consult Internally On CI/CD Improvements For Team - Sensitive info - Design - Implement GitHubActions - AWS Lambda - Testing
Here the situation was interesting. There was a team that was staffed with analysts and was still about 3-months away from hiring their own data engineer. (Visa issues really are the bane of hiring.) So in order to help them and to allow the new engineer to be productive near their starting date myself and another engineer from my team were tasked with figuring out what exactly they needed and to come up with a plan to meet their requirements. This project was a ton of fun. Oh and they were working with super sensitive employee info.
Situation: Design an easy-to-use, repeatable, and standard way to deploy new code to existing AWS lambda functions
Task: Gather requirements via user interviews, balance requirement of shipping a solution quickly vs. using an existing internal framework
Action: Delivered a CI/CD solution based on Github Actions that would allow the team to build, test, and deploy their code to AWS Lambda with confidence
Result: Team was able to move from local testing and manual deployment to production to an automated one that is documented and managed. This effort increased the number of iterations an analyst could do on a query by 100%.
All Together: In collaboration with a teammate we designed, implemented, deployed, and documented a CI/CD pipeline using Github Actions for a team of Data Analysts in a data agnostic and privacy preserving way never being privy to any sensitive data. We gathered requirements via stakeholder interviews and we evaluated various solutions ultimately deciding to go with GitHub Actions implementing a simple yet powerful pipeline to allow the team to iterate quicker and test with confidence before deploying their new code to run as AWS Lambdas.
All the parts that don’t make it into the bullet-points (I am trying to keep them brief) I will take note of when being interviewed so I can answer questions like: “Tell me about the this here on your resume where you mention you implemented a CI/CD pipeline for this team…”. It is a good habit before an interview to review one’s resume to be ready for questions about previous roles and tasks.
The final product would look a bit like this:
Migration from Redshift to BigQuery - migrate key dimension table - optimize schema for bq - airflow dags - QA
Sitation: The company has decided to shift from Redshift on AWS to BigQuery on GCP
Task: I was tasked with planning and executing the migration of a key dimension table to BigQuery
Action: Expanded the scope of the effort to reduce some techincal debt and due to BigQuery’s pricing model determined the needful columns to migrate, migrated the new table and newly optimized schema onto BigQuery, QA’d the data against the old table and reports
Result: Expedited the migration that other teams were engaged in by promptly delivering the migrated dimension table used in 100s of reports onto BigQuery with a newly optimized schema with no reporting downtime
All Together: While working to help achieve the goal transitioning from AWS Redshift as our Datawarehouse platform of choice to BigQuery on GCP, an effort that took many months to validate and coordinate, I was primarily responsible to migrate the most important dimension table to our model used by 100s of reports and stakeholders every day. Taking into account BigQuery’s new pricing model and the timeline for the migration I took the opportunity to work with stakeholders within and without the team to determine columns still relevant and those that could be set to null for the migration to save on storage costs. Also conducted analysis to determine the most efficient partitioning and clustering columns.
Migration from Redshift to BigQuery - migrate key dimension table - optimize schema for bq - airflow dags - qa
Sitation: The company has decided to shift from Redshift on AWS to BigQuery on GCP
Task: I was tasked with planning and executing the migration of a key dimension table to BigQuery
Action: Expanded the scope of the effort to reduce some techincal debt and due to BigQuery’s pricing model determined the needful columns to migrate, migrated the new table and newly optimized schema onto BigQuery, QA’d the data against the old table and reports
Result: Expedited the migration that other teams were engaged in by promptly delivering the migrated dimension table used in 100s of reports onto BigQuery with a newly optimized schema with no reporting downtime
All Together: While working to help achieve the goal transitioning from AWS Redshift as our Datawarehouse platform of choice to BigQuery on GCP, an effort that took many months to validate and coordinate, I was primarily responsible to migrate the most important dimension table to our model used by 100s of reports and stakeholders every day. Taking into account BigQuery’s new pricing model and the timeline for the migration I took the opportunity to work with stakeholders within and without the team to determine columns still relevant and those that could be set to null for the migration to save on storage costs. Also conducted analysis to determine the most efficient partitioning and clustering columns.
BigQuery Cloud Cost Optimization - regularly read the updates the BigQuery team post about best practices and pricing changes - identifying best practices to reduce cost - reading query plans and adjusting as appropriate to have the most cost efficient queries - making tradeoffs between cost and speed
Situation: BigQuery is constantly improving and adding new features and sometimes iterates on pricing
Task: Be aware of new features and pricing changes, where appropriate, enforce partition filters on partitioned tables, in code-review create a framework for evaluating SQL queries by understanding the query plan and looking for anti-patterns
Action: Make the team and other engineers company wide aware of feature changes and pricing changes that would affect them
Result: A strong team-wide effort to understand costs and design a performant, cost-effective Datawarehouse on BigQuery
All Together: Helped to bring a culture of seeking the most performant, most cost effective approaches to datawarehousing on BigQuery by constantly keeping up-to-date on the latest developments of BigQuery in both features and pricing and by performing rigorous code reviews looking to understand the query-plans of SQL queries and looking for anti-patterns, enforcing partition filtering where applicable, and other techniques
Hiring / Interviewing - Writing coding challenges for SQL and Python from real-life scenarios and problems I’ve faced in my day-to-day - Interviewing senior engineers and managers - Running final-round interviews
Sitation: Hiring in tech is broken. I am doing my part to make it better.
Task: Encourage strong engineering talent to join DeliveryHero; hold interesting and respectful interviews; Migrate take-home project tasks to online programming excercises; expand online excercises wtih tasks, problems that come up in everyday projects
Action: Active on LinkedIn and other platforms to encourage talent to apply to DeliveryHero; improved online coding challenges by ensuring that the challenges were based in real-world, applicable scenarios and not HackerRank or Leet Code style exercises that are not applicable to the work being done; periodically expand the tasks with sitations that I have personally solved that I think could make for interesting interview questions
Result: Recieved feedback from interviews I’ve conducted with potential hires praising the experience as one where they learned and were treated with respect and were given the space and time to be comfortable and perform their best. Turnaround time for new hires fell from ~1-2 months to 3-weeks.
It’s good to add numbers to the results section. Numbers speak louder than words sometimes. It’s far more effective to say “reduced mean time to deployment of the pipeline by 50%, ~5 mintues to ~55 seconds” than to say: “made the CI/CD system faster”
All Together: Improved the interviewing experience by removing the over emphasis on esoteric algorithmic coding challenges often seen on HackerRank or Leet Code that have ruined hiring in tech and for Data Engineers. Modernized the take-home engineering tasks we used to give and adapted them to fit 30-minute coding challenges in Codility for use in the 1-hour interviews. Improved the wording of the challenges for candidates whose native language was not English. Ensured that the tasks were applicable to data engineering and were interesting. Created an environment where interviews are a safe place for candidates to show their skills and obtain an understanding of the values of DeliverHero and what it would be like to work together.
Formalize Dev PR process and Documentation - champion for it - add docs to the workflow
Sitation: Documenation and best practices for code review and authoring can be found in many places, unclear what to do in code reviews
Task: Document best practices in code-review and define a “contract” for both code authors and code reviewers
Action: Codified documenation of requirements best practices as a checklist with context and documentation built-in for code authors and code reviewers and added it to the Github PR workflow for developers and reviewers to see
Result: One aspect of the requirements for code authors and reviwers was to check for documentation – because of this checklist documentation 100% of significant, core code paths have code comments and associated confluence documentation for additional reference
All Together: After writing and implementing code authoring best practices (code style, local testing and testing in staging before submitting, requiring code comments and documentation) for both authors and reviewers documentation for critical infrastructure and core models is > 95%
The above took a long, long time. But this is the approach one should take for all their previous roles. But I would refrain from spending a lot of time on past roles as one’s current role is likely the most applicable to the role you’re interviewing for. So spend the most time here.
Mentoring / Training - git training video link - Noticing many engineers, especially juniors, struggling with git I created a youtube presentation on my own time going through a simple code change and the associated git workflow - created additional documentation on using Git at the commandline for both Linux, MacOS, and Windows with WSL
Sitation: Noticing in Slack channels and elswhere engineers and managers struggling with git, especially at the commandline
Task: Be able to contribute to the mono-repo using git and Github
Action: On my own time, created a Git tutorial video going through a canonical SQL change and the associated git workflow
Result: Many engineers benefitted from the tutorial video and the subsequent live-training, the video and documentation are still referenced internally today
All Together: In response to noticing many engineers struggling with Git created internal, detailed written and video guides and a live training session for how to best use Git and Github in the context of the internal mono-repo
My previous roles are not that interesting. I will finish that outside of this article as I think what has been demonstrated shows how one could set about systematically improving their resume using facts and figures and action statements. This approach will leave you with more confidence when interviewing and will elevate your resume from something you give out to be parsed by a machine into a tool that you can use to aid you when talking about your skills and what you bring-to-the-table.
Put down your education starting from University/College and upwards. If you’ve only ever completed High-School or even didn’t finish just put what you feel comfortable putting. In our industry some roles will pre-screen applicants into the “no” pile for want of a unviversity degree (which I think is stupid and short sighted) but with experience one can be very productive and capable degree or not.
Ahh “Acronym Bingo”. This is a last hurrah of sorts in case the work above hasn’t quite appeased the bots that scan your resume. Here is the kicker to this:
ONLY LIST TECHNOLOGIES THAT YOU ARE COMFORTABLE WITH, THAT YOU HAVE STRENGTHS WITH, THAT YOU’D BE OKAY GETTING QUESTIONS ABOUT
The keyword here is comfort. I am not the world’s best SQL developer or even Python engineer. But I have had enough experience with both that I could talk for hours about it. It’s fun. In fact, some of my most enjoyable interviews were when I was geeking out with an interviewer about various databases or coming up with alternatives to solve some query prompt in SQL.
Putting every fad technology under the sun here or every bit of tech that a data engineer might be exposed to is suspect. Nobody is an expert in everything or has seen all the technologies.
If I were to see a senior engineer with about my amount of experience in the job (~3 years) I’d expect to see:
SQL; Python; One or more of the clouds: Amazon Web Services (AWS), Google Cloud Platform (GCP), Azure; Airflow; ETL; ELT; Kafka; S3; PySpark; Source-Control (Git); Shell Scripting; Linux.
Write out “Amazon Web Services” in case the scanning system is too dumb to know that AWS = “Amazon Web Services” and then also put the acronymn in parentheses.
Of course as of 2023 there’s far more tech out there that I haven’t listed because I am not really confident in them yet:
DBT, Terraform, Kubernetes, Golang, basically any of the litany of really cool and capable Apache Foundation projects for Data Engineering, etc. For example I’ve written Golang but only really for personal tools and projects.
This could be it’s own article and it likely will be. But let me talk about it a bit here. I think certifications are amazing for folks just getting into the industry. It means that someone taking a chance on you is that much more confident that you’ll be able to do straightforward tasks than someone simply out of university.
For example someone certfied as a RedHat (Linux) Certified Engineer would be able to add users, stand up servers, and maintain them because the certification tested for that and it’s assumed the person who got the cert is capable of doing all that. As someone who might view this resume vs. someone without it I would prefer the certification because if I was hiring for a Redhat Systems Administrator it’s directly applicable to the role I am looking for.
I always just put: “References available upon request”. But what I am thinking of now is to ask former colleagues and managers or even current teammates to provide a quote or two about what it is like working with me. Personally, that might be a bit too much self-promotion but I think it could be cool.
If you do go this route follow this:
“Quote” 2-3 sentences. Name of the person providing the reference/quote, first and last initial, and their title, and where they work.
Example:
"I worked with Alex at DeliveryHero and he was always happy and cheerful and always took time to help me..."
John S. Senior Data Engineer - AdTech - DeliverHeroThere’s another benefit to listing references and it speaks to the nepotism that pervades this industry. It seems that in order to get a job these days it’s less about what you know and more about who you know.
I think there’s some truth to that. Two of the most capable engineers on my current team I referred, vetted, and vouched for and fought for. Because I knew them and the immense value they could bring to the team I referred them I had a vested interest in their being hired. This is not something one gets when simply applying to a role as one of 1000 resumes. Stay tuned for anothe post about effective networking. (Spoiler: One has to be genuine about it and not transactional – transactional professional relationships can be seen a mile away and won’t get you very far.)
So consider adding endorsement quotes to your “references” section especially if those endorsements come from someone at the company you are applying to or are significant / influential people in the industry.
Find the finished resume here.
I hope you found this useful. Especially the bit about doing some internal reflection about what you’re getting out of your current role besides a salary. The industry is moving quickly. We cannot afford to stand idly by and not keep up, we need to be constantly learning and improving.
With that said a good prompt and some bullet points ChatGPT could probably write a compelling resume and you’d probably not even have to read all of this. But I am glad you did. Even if you go the route of using ChatGPT or some other resume-writing LLM you can use the above points to fine-tune it and make it really stand out.
I’ll follow this article up with one where I try to get ChatGPT to write me a better resume.
Copyright Alex Narayan - 2024