DEV Community: Walt Schlender

It’s hard to get job offers at tech companies right out of school. What's going on?

Walt Schlender — Tue, 27 Jul 2021 12:19:13 +0000

I used to work in engineering and data science at a tech recruiting company.

We had a hard time placing new grads.

In this article I want to describe what tech hiring is like. If you’re just starting your career I hope this article can help you come up with a strategy to stand out from the competition and get the job you want.

It's hard for new grads to get their first job

According to data I've seen, candidates who get a lot of job offers have a mix of experience (preferably from a famous company or school), the right skillset and the right salary expectations. In that order.

Experience is particularly important.

The data I've seen suggests that there's a 3 year career-start dead zone -- you may be an amazing engineer but during those first years you'll have to knock on a lot of doors to get your interviews.

In an industry "short on talent" where companies happily pay head-hunting firms $20,000 for introductions, why do new engineers struggle to even set up brief meetings?

A lot of new engineers are talented, passionate and filled with potential. They're less expensive and fairly abundant.

You'd think companies would be taking advantage of this talent pool but many of them aren't.

What's going on?

I believe that companies are being careful. Making ANY hire is expensive. A bad hire can be catastrophic.

Let me describe what the interview process is like from the employer side.

How does interviewing work inside companies?

In most companies finding candidates is the job of the HR / recruiting department or external recruiters.

Companies have limited resources so out of the 12 applications they get for a position they'll only likely screen a couple.

A basic code screen requires ~2 hours from an engineer whose salary after taxes and benefits runs about $100 / hr. Time spent interviewing is time NOT spent coding. When you factor in time for context switching and getting lunch an interview takes about half of an engineers' day.

Recruiters usually don't have engineering backgrounds themselves but they're the ones who take the first look at your resume. They tend to work by keying in on keywords.

Because of this a resume focused on 'Rust', or 'Clojure' skills will get weeded out. While an engineer looking at this resume would see "rockstar coder with an interest in functional languages", a recruiter only sees "not a node engineer".

A company usually spends one to two full days interviewing a candidate.

The process is elaborate because the stakes are high. While rejecting a candidate is painful, hiring the wrong candidate is disastrous.

Hiring the wrong person is really bad

The hiring process itself is expensive but the real costs come after the hire.

Closing costs

The company spends a pile of money before the new hire even walks in the door.

If a recruiting firm sourced the candidate the company has to pay a finders fee - typically ~20000$ (15% - 20% of a first years’ salary).

Then there are all the costs associated with hiring. Payroll, IT, etc...

Ramp-up

Like most jobs, it takes a while for a new employee to become valuable. With engineers I've it takes about 3 months.

During this time the engineer needs a lot of hand holding.

An experienced engineer will need help with the quirks of a company's system, but a new engineer will need help with system setup and basics.

Code quality

A newer engineer also requires more babysitting.

They may not have mastered architectural patterns. They may not have experience with testing and code reviews.

The company is exposing itself to a lot of risk if code ships with bugs or the codebase gets crufty.

Older engineers write bad code too but if they worked for Apple the odds are that they have fewer holes in their coding skills -- or at least that's how the thinking goes.

Nobody likes firing people + so hiring mistakes don't get fixed

It's easy to hire someone it's really really hard to fire them.

There have been a few times in my career where I worked with someone who obviously should have gotten fired.

One person was showing up to work drunk and trying to write code. Another really took advantage of the 'unlimited vacation' policy by basically never showing up to work.

In each case it took months for the management to finally admit that a mistake had been made.

Firing someone after investing team time and resources sucks.

If a person has a strong track record the likelihood you'll have to undo your hiring mistake drops.

Newer engineers are up against a lot... but we all got started somehow

I don't want to leave you with the idea that getting work in tech is impossible. Lots of people have broken into the industry even without CS-degrees.

I know this because this was actually my path.

When I graduated I had a lot of coding skills but my degree was in economics.

My own story involved building up a portfolio by doing small-time contracts for clients with small budgets. I then used that portfolio to get bigger more prestigious clients.

Over the years I have learned that getting hired takes strategy based on a realistic assessment of your strengths and weaknesses.

Not all 'employers' are the same.

Some companies are in high demand. Everybody wants to work with them so they're inundated with resumes from rockstars. Other companies need people but don't know how to find them. They'll hire the first good engineer who walks through their door.

Little changes in a resume (like emphasizing your Node skills over your older .Net skills when applying to newer companies, or changing the market you're searching for a job in) are the difference between 20 job offers in a week and crickets.

A little bit of thought about strategy can really help.

I thought it would be useful to end this article with some strategy recommendations.

I hope some of these approaches are helpful to you.

Strategies for getting in the door when you have no work history

To begin with, set realistic expectations - expect the hunt can be hard.

If you're sending out lots of resumes and not hearing back know that this happens to lots of candidates -- it's fairly normal so don't lose hope. All it takes is one company to say yes to hiring to get things rolling.

Sending out resumes can work but in my experience it's better to see if you can bypass HR by talking directly to an engineering manager.

The best way I know of to do this is to attend tech-focused meetups and events. Search for ruby or node or whatever tech meetups and go to the events. After the event employers get up and see if anyone is looking for work. Talk about some of the stuff you've done or things you're passionate about and you'll likely get invited to come in for an interview.

Another strategy you can use is to work with an employer in a contract (often contract-to-hire) role.

As we talked about earlier, the big things that are keeping the employer from working with you are cost and risk. If an employer doesn't have to actually commit to hiring someone full time they're way more likely to take a chance on someone with a bit less experience.

You can find contract work at events, on contract-centric work sites like Upwork. Smaller companies in particular like to work with contractors. This is how I got started and I highly recommend this approach.

Sometimes you really DO want a role in a hard-to-get-into (popular) big tech company and when that’s the case you need to make your resume stand out.

As someone who can't rely on work history to get in the door, you’ll need something else to make your resume stand out.

Some things that you can do ~

Niche ~ get specific about who you want to work for and bring more to the table than just programming skills.

Most candidates are telling employers, ‘you need Java? I have Java’ but the employer really actually needs ‘Java’ to solve educational or fashion or recruiting problems…

If you can bring a passion for the space along with the skills you have a much greater chance of getting an interview.

Demonstrate work the employer is looking for ~ a portfolio with relevant projects is pretty awesome here.

One hack I've seen done is to give yourself a homework assignment. Research a company you'd like to work for and put together a mini project (as if they had hired you). Send the project to them as part of your application.

Demonstrate high quality engineering skills.

Most new grads are missing professional engineering experience. They don't know about testing of version control workflows. They have nothing a potential employer can look at.

The solution? Work on open source projects, go to hack-a-thons, get involved in the engineering community.

All of these things can help you get interviews... once you get interviews all you have to do is pass them.

If you're interested in learning about passing coding interviews I put together a 50 minute Skillshare course about coding interview techniques. It's free for dev.to readers until August 30th. Check it out.

I hope this article has been useful for you and that you have some new ideas for strategies you can use in your job hunt. Good luck and may you find exactly the work you’re looking for.

My first two deep learning projects

Walt Schlender — Wed, 12 Jun 2019 08:16:36 +0000

Last week I started taking the fast.ai Practical Machine Learning for Coders course.

I learned how to build a state-of-the art image classifier in about an hour which was amazing! Then I started trying to apply what I learned to my own projects. Here’s what happened.

Project #1 - YouTube thumbnail grader

My first project attempt was to build a classifier that could judge the quality of YouTube video thumbnails.

Something like this would be so cool! Imagine being able to plug in a thumbnail for your newly created YouTube video and get back recommendations on how you could improve it.

As a first step I tried to build a classifier that could recognize good thumbnails.

I collected together a bunch of YouTube thumbnails and estimated how many views per day the videos were getting. I labeled the videos as getting ‘less than 100 views per day’ and ‘more than 100 views per day’ and tried to see how accurate the resulting model was.

I had a feeling this test wouldn’t work very well… there are a lot of factors besides the thumbnail that influence view count (# of followers + fame of the YouTuber) and I wasn’t wrong. The performance was dismal.

The model thought that this thumbnail would get less than 100 views.

This makes sense though… It’s probably the case that thumbnails (while important) don’t make a HUGE difference in view counts.

Honestly I bet you I could make a really good model for predicting view counts. Just regress view counts against subscribers and you’re good. That’s why the YouTube pros alwasy say ‘Like and subscribe’!

So the project failed?

Absolutely not! The whole point of the work (at this stage) is to figure out what machine learning is capable of + go through the steps required to train a model. I definitely did all that.

This project taught me the meachanics of deep leraning + I started to get a more intuitive feel for what kinds of problems a model would succeed or struggle with… in this case the model struggled because it didn’t have the information it would need to make a good judgement.

Project #2 - Chinese language detector

My second attempt invovled building a model that could listen to people speaking and classify the sound as either ‘English’ or ‘Chinese’.

To do this I took audio files (which store audio data as “time-series” amplitudes-over-time)… and converted their amplitude over time data into spectrograms (pictures of the frequencies present in sound over time).

This —>

Got converted to this —>

The hypothesis was that the machine learning model would be able to learn the differences between the English and Chinese audio files and would be able to pick out which one was which…

The model’s performance on this project was interesting. It did fairly well at classifying files that were part of the initial training set.

But when it actually came to predicting samples of me speaking Mandarin it was pretty horrible.

Basically the training and testing of this model were… frustrating.

There are so many variables that could cause problems for a model like this.

The original audio file encoding
Whether or not the audio files are clean
The processing of the audio files into spectrograms
What kinds of features the model is picking up on — (do I have specific vocal characteristics that will mean the model always picks English for me???)

The list goes on.

After doing a bit of research I learned that CNNs (convolutional nerual networks) may not be the best kind of network to apply to a problem like this since the data in the pictures is more repetative than fixed.

I found some researchers who succeeded in solving this problem using RNNs (a neural network where the network bases it’s prediction on data that came from a previous piece of data) instead of CNNs. These networks are good for these kinds of tasks because the network would say… ‘given that I just saw this shape the next shape should be… blah if this were english’.

So I think solving this problem is a bit beyond my present skill level, but I’m not giving up! I’ll put this one on the back burner and pick it up again when I know a little more.

So what are some things I learned from this project?

A lot about signal processing and the Fast Fourire Transform (the most amazing and cool math tool out there)
That CNNs struggle when the pattern they’re being asked to classify isn’t clear and obvious
That I would probably learn faster if I worked with datasets and problems where I can see other people’s answers (maybe contests?)

Taking a step back, why learn about this stuff?

Honestly, after a week of working with this tech I am blown away!

I can’t yet see exactly how this stuff will change the world but I have no doubt that it will change… everything.

My bet is that with a little more experience I’ll be able to build some useful tools. We’ll see. It’s definitely interesting so far and I’ll keep you posted as I learn more.

My 2018 year in review

Walt Schlender — Fri, 01 Feb 2019 08:16:36 +0000

In 2018 I wrote some articles, leveled up my design skills, built some android apps and used data to study some oneline marketplaces. I learned a lot about entrepreneurship. Hope you enjoy!

2018: the writing year

Looking back on 2018 the theme was less code… more writing.

I started off the year posting a couple of articles I wrote… then shifted back to writing software… then went back to writing articles.

My understanding of the entrepreneural process flipped from -

Come up with an idea
Build it
Find customers…

To -

Find customers
Test the market with mini products (content / landing pages)
Build a bigger product if there is a market

Here’s some highlights from the things I tried in 2018.

I gave my wife a coding puzzle for Valentine’s day and I challenged Reddit to solve it

I started off my marketing tests by writing an article for Medium.

It was Valentine’s day of 2018 and the WHOLE WORLD was obsessed with Etherum.

I had just completed a job where I wrote some code that stored some data into the Ethereum block chain.

One morning my wife asked me how ethereum worked and I had a brainwave!

I stored $100 of ethereum on a secret address in the blockchain and created a puzzle for my wife. If she could solve it she’d get the money.

I sent her the puzzle on Valentine’s Day then wrote an article so that other people could try the puzzle out too!

Valentines day puzzle

Results

I shared the article on Reddit and posted it to Hacker News.

This was the first time I had ever shared something I’d written on the internet so I had NO IDEA what I was doing. I ended up violating two reddit communities’ terms of service and hitting the hacker news spam filter.

Still, that article got about ~3000 views and proved to me that writing articles is a good way to reach people.

I learned so much with this first article! Some highlights include:

That stories are actually little products - check out Donna Lichaw

That there is an art to creating stories that go viral - Karen X Chang is incredible!

That it’s really challenging (and important) to learn to write clearly - check out On Writing Well

This was only a small victory, but it was very important. I got much better at writing but that was later in the year. First I had to learn about design.

I leveled up my design skills

If you want to build your own products design is SO important.

The design process involves getting really clear on who a product is for, what problem the product is solving, how the product will solve that problem and why life will be better as a result.

Good designers answer these questions and then come up with blueprints that can be made into products.

I’m fortunate because my wife is a designer. I asked her to teach me her process and she didn’t disappoint.

I recommend checking out her design challenges and downloading her design cheat sheet. I learned a ton and I’ve used what I’ve learned to build mockups and landing pages.

Here is a screenshot of the first app I designed after she taught me her process.

Designs for my 30-day challenge tracking app

I built (and launched) the app I designed… in 3 days

Since I had a design, I set myself a challenge… how quickly could I throw together an app? Three days it turns out.

Three days is pretty quick for an engineering project (even a simple one). The build was quick because I used Firebase and React Native. These technologies are very useful for building quick prototypes. If you’re building MVPs this is the stack I would recommend.

Results

I launched the app on the Google Play app store.

I got 800 downloads over the month the app was live. A single user purchased the full version so the project was profitable! I made $3.

After a month I took the app off the app store to try other ideas. The app needed work and I wanted to work on other things.

I wrote a popular article for the freeCodeCamp publication

I was looking for a way to drive US traffic to my projects without spending a lot of money on Facebook ads.

I had read that people who run blogs and publications are always looking for new high-quality content to their audiences.

I put together an experiment… write an article on a subject I’m very familiar with and see if I could get published.

Results

My article on landing coding gigs was accepted to the freeCodeCamp publication and distributed the the ‘Freelancing’ medium topic.

I was surprised at how well my article did. It received over 2000 claps in the first week and showed up all over the internet.

The article was published in August and continues to get viewed by ~200 people per day.

I added links to my blog and email list and twitter to the article and these have kickstarted a personal email list.

Stats for my article on how to land a coding job with very little coding experience

I learned that -

Guest posting is a very effective way to reach large audiences
That publications are easy to approach - you’re helping them when you write a good article
That the internet is designed to spread articles… so articles (and videos, pictures etc…) are a good way to draw attention that can be diverted towards bigger products

This was a big victory. I have always struggled with driving web traffic. My android applications were successful because google delivered ~30 new customers per day. I have been looking for a way to get that stream of customers for web products. This seems like it works.

I ran 2 landing page tests ~40 people signed up for my pre-launch list in 1 week

I kicked off September by taking another design class. I felt that I needed to be able to create pretty designs to sell my ideas.

The class was super helpful. I became pretty comfortable with Adobe XD and was able to produce some pretty nice looking landing pages.

Freelance Map landing page test

Product testing 101 class landing page

Freelance Map

My first landing page test was for FreelanceMap.

The pain I was addressing came from personal experience.

It can be lonely to work alone. With FreelanceMap you could use an app to find other local freelancers.

To test this value prop I built a landing page and drove traffic to it using Facebook ads.

The main landing page leads to a pre-release signup form.

‘Get the app’ - call to action

People could sign up to hear when the app was ready.

Results

I ran the ad for 1 week and 39 people signed up to be notified when the app is ready.

1 week of running ads at $5 / day gave me 39 subscribers

I felt like the test went really well.

The experiment design was pretty good but not perfect.

First, the ads were run in India because traffic from there is cheaper. I still wonder if US traffic would have converted differently.

Second, I never asked people to pay for anything. After running the test, I was left wondering if building the product would make me any money.

I didn’t want to build something that people wouldn’t pay for, so I tried running another test.

Product Testing 101

Product Testing 101 pitched a class about building landing page tests.

My experiment design was a bit different for this test.

The landing page was very text heavy. I tried very hard to convince people that the class was worth paying for.

Down at the bottom of the landing page I asked people for money and I installed a plugin that let me track how far down the page people read.

Asking people for money

When you come up with a product idea there’s a lot of risk around whether you’ve found a real audience whose pain you understand.

I personally feel that the value proposition for this product was more contrived.

Apparently Facebook users agree with me. You can see that the clickthrough rate for Product Testing 101 are half the clickthrough rates for FreelanceMap.

.7% CTR for product Testing 101 vs 1.52% CTR for FreelanceMap

Results

Nobody clicked through to the pre-release list signup page.

That said, I considered the test a success.

Scroll tracking showed that a lot of people read my entire pitch.

Most people bounced when they got to the pricing CTA.

I felt that this test design gave me a clear answer — this product pitch (addressed to this audience) wasn’t strong enough for people to be willing to pay for it.

Overall I felt that these landing page tests really worked. Tests like these are a really good when you have found an audience, understand their pain, and want to know if you can reach more of them.

These tests don’t work so well when you’re not clear on who your audience is or the pains you hope to address. If you want to learn about audiences I think doing a bit of research yields better results. After running the landing page tests I tried a research based approach… read on to hear how it went.

I did some research and wrote the most ‘clapped’ article for October on freeCodeCamp

I found myself looking for a way to identify an audience and understand where they needed help.

I designed an experiment where I picked an audience I knew a lot about (new programmers) and I spent some time reading forums to get a sense of what kinds of problems they were struggling with.

I combined my notes with my own personal experience and came up with an article about the mental side of learning to program.

That article turned out to be really helpful. People really seem to like it.

As I write this article that article has gotten over 25K claps and was viewed over 9,000 times in a day. It also ranks first in google for the keyword ‘struggling developer’.

My article was the top article for October on freeCodeCamp and it ranks first in a google search for ‘Struggling Developer’

That article continues to generate traffic and followers for me.

To me it’s validation that it’s possible to identify a market hole through research.

I’ve come to believe that research is a very valuable tool for getting clear about who you’re building something for and what they need help with.

Research helps you move from guesses to facts… but I find it to be really difficult. What is an audience and how do you find them online? When you read a forum thread, how do you know that the problems your reading about are widely experienced?

I thought that software could make the research process easier. Also, my background is in data engineering + I think I really was looking for something to code.

I built a couple of data tools. They’re very technically ‘cool’ but they didn’t turn out to be very useful. Still… they’re worth mentioning.

The Reddit research tool

When I began researching audiences, I turned to Reddit.

Reddit divides communities by ‘subreddit’. Different communities have different focuses, cultures and interests. I thought it would be cool to take a subreddit, extract all the keywords from the forum posts, tally up recurring keywords and see what gets talked about most.

I built an interface where you can search for a keyword and see all the subreddits where that keyword gets mentioned. You can also search for a subreddit and see what keywords get mentioned there.

My reddit research tool makes it really easy to x-ray a subreddit to get a sense of what topics appear in a subreddit

After I had built this tool I started using it.

The tool is actually pretty useful. It lets you use keywords to quickly hone in on subreddits with people talking about a certain topic. I found a lot of interesting subreddits I never knew about using this tool.

That said, I haven’t had a lot of success coming up with business ideas by trawling reddit forums. Reddit has a lot of students so I kept finding myself running into pains associated with academic work or finding jobs. I’m not saying this approach can’t work… but it didn’t work for me.

Still, I wasn’t ready to give up on quantitative approaches just yet.

The Android market research tool

While doing research I happened to come across a company that uses Amazon data to find product opportunities on Amazon.

Jungle Scout uses Amazon’s ‘Sales Rank’ metric to estimate daily product sales for every product in Amazon’s catalog. By combining this sale estimate with product review data people who use Jungle Scout can find products that sell well despite needing to be improved.

I figured this approach might work with the Google Play Store. If I could find apps that were getting downloaded a lot but had poor reviews maybe I could build something better.

I ended up scraping the google play store to collect about ~60,000 app listings.

It turned out it was pretty easy to see what kinds of apps get the most downloads…

Games!

A bit of data crunching led me to some niches I thought I could build apps for.

I thought I had found a niche in the ‘Name Art’ market

I thought I’d had found a market opportunity when I came across apps for creating Name Art.

These apps are little drawing programs that kids use to trick-out their name and send them to their friends.

The apps are all pretty low quality and are LOADED with ads.

I thought I could whip up a similar app and so I got to coding. This is the result.

My social signs app

I didn’t get many downloads after releasing this app.

A bit of research revealed that the reviews that I had been basing my decisions off of were fake.

Moreover, I building a product by looking at competitors just didn’t work very well for me. I kept finding new competitors… and since I had no connection to my customer base I didn’t really understand why they would or would not want an app like this.

Maybe a ‘copy the competition’ approach could work but I think it seems easier to go study the customers. After all, if they have a problem and don’t know about the competition then the competition is irrelevant.

2019

So that brings me to 2019.

As I write this, I’m still searching for my product-building approach.

No doubt about it, this journey has not been easy. Each success was accompanied by A LOT of failure… and even things that worked (like writing articles) weren’t always very repeatable.

I’m still optimistic though.

Looking back I can see how far I’ve come. I started 2018 with a very different entrepreneurial world-view. My understanding keeps getting more complex and nuanced.

I think it’s only a matter of time before I settle on a product-building approach that works for me.

Until then I just focus getting up each day and putting in the work.