from Guide to the Job Hunt on Mar 26, 2023
Why (and why not) work at Apple
As of time of writing, I've spent just under a year at Apple as a full-time research scientist. I love the team, the work, and the culture within our group; needless to say, I'm very happy with my decision to come here for full-time. However, there are company-wide culture quirks that impact your day-to-day.
This post is written for anyone considering a job at Apple, especially for those with an avid interest in the working environment. To be specific, this post is written for anyone that would ask "How is it like at Apple?" To assess the environment, I organize this post around pros and cons, influenced by infrastructure, leadership, and culture.
As I discussed in How to make big decisions., discussions for major decisions should be broken up into a) criteria and b) rankings. For each of the major points below, I'll a) lead with the criteria and b) present information that will help you rank Apple for that criteria.
Long-term sustainability is the main pro.
The main pros make Apple a great place to work at long-term, sustainably. In short, you can be a healthy, productive, and reasonably intense-but-sane employee.
Pro #1. Mature hardware tooling. This is really where vertical integration shines: Profiling on Apple silicon is a breeze, as there are tools both for profiling on the current device as well as remote clusters of devices — both simulated and actual performance.
- In general, tooling is very mature. All internal services are held to a high standard, so internal APIs and services are trustworthy. By default, services are provided to internally-built applications, distributed through a private app store.
- Impact: Measuring on-device resource usage on a variety of different chips and devices is simple as a few API calls. This makes it relatively easy to profile performance across scenarios and devices.
- Comparison: Testing on actual hardware was a highly manual process at Meta — procure your own devices, set them up, and run tests manually. If you were lucky, you could partner with hardware testing teams, but the process was still just as manual for them. Furthermore, Meta's internal services are always evolving, breaking, and failing, as we discussed in Why (and why not) work at Meta — a far cry from Apple's standard for its internal tooling.
Pro #2. Robust and flexible compute infrastructure. To my knowledge, teams are granted pools of GPU hours, and the team may use these freely. You can persist a cheap debugging node for 14 days at a time, launch test runs en masse, or launch a single expensive multi-node run. All these resources are dedicated, non-pre-emptible.
- Impact: This is a huge productivity boost, as compute infrastructure is never the bottleneck when it comes to executing a project. Granted, the compute allocation may need some planning around, but working with the infrastructure — e.g., launching jobs, setting up debug nodes, or babysitting runs — is a breeze.
Comparison: In short, this is the best compute infrastructure across academia and industry
- This is a huge improvement over academia, where each lab has its own compute cluster. Fortunately, some clusters use slurm for job management, but with a small cluster and without usage throttles, a lot of human interaction is needed to get even a debugging node. There weren't enough nodes to run experiments at scale. Using AWS was furthermore a pain, as obtaining service quotas was painful and slow.
- Meta's infrastructure only allowed a single debugging node per team, so in a similar spirit, human interaction was needed to get resources for debugging. Fortunately, the compute cluster was large enough that for the most part, human interaction was not needed for running jobs at scale. Unfortunately though, nodes in the cluster were prone to crashes, and the job submission API was constantly changing, making it necessary to frequently update or fix the job submission script itself.
- Tesla's compute infrastructure was the closest in ease-of-use, as discussed in Why (and why not) work at Tesla. However, Apple's global pool of dedicated GPU nodes is far larger than Tesla's, making large-scale experiments easier to run without coordination.
Pro #3. Incentivized to be healthy. At a surface level, Apple does have the familiar tech company perks — free food, free swag, and others. However, there's a common gatekeeper across all of these "free" perks: You have to be healthy to earn it.
- Meals are subsidized but still cost a reasonable amount, meaning your food is automatically portion-controlled. The price isn't enough to drain your wallet but enough to stop you from eating uncontrollably.
- There are no free snacks or free coffee1. The food in cafeterias are also decently healthy. This means that the food you do consume in the office is almost always healthy.
- There are probably 5-6 opportunities for quality company swag per year, but to earn these, you must hit certain health goals for a 30-day period — ranging from eating fruits and vegetables, to hitting daily calorie targets, to meditating.
- Company insurance Aetna incentivizes you to hit certain fitness goals with approximately ~$10 in gift cards per month, via an app called Attain.
- Impact: Overall, this makes Apple a long-term healthier option, without the common temptations of free snacks, desserts, and food in general. Instead, the company uniquely incentivizes you to reach fitness goals with company swag. In short, nothing is free, but the cost to earn is to simply be healthy.
- Comparison: We could draw a line from stingy to generous, staring with Tesla at one end to Meta at the other. Tesla subsidizes meals technically, but instead of lowering your dish's price tag, they instead guarantee each food truck's daily earnings. Meta has free food — all the snacks, desserts, and ribs you could want. Granted, I loved this as a student, but long-term, this is definitely unsustainable for my health. Meta does provide a "fitness" budget per half, but this is less effective than all the temptations in the office.
Pro #4. Sustainably intense team. In short, there are long hours, but these aren't expected. The team is capable of raising intensity for key deadlines, but these deadlines are planned in advance and our efforts are recognized. In short, the team is intense-capable without burning out.
- Team Slack channels will often be buzzing late into the night, as late as midnight. Posting at 9 pm is not uncommon, and teammates will often respond or react immediately. We've spent 12-hour workdays at the office as necessary. Even though most folks put in long hours, this isn't an expectation, making the extra hours relatively stress-free. We're simply invested in the work.
- Impact: This confers many of the benefits of an intense team — you can get answers at any hour of the day, when it's crunch time. If the need arises, the team will put in the work and hours to make deadlines. This is especially helpful for critical product-facing work as well as papers.
- Comparison: Tesla's culture doesn't "allow" burnout. You can take and maximize vacation days, but you're out of luck if no vacation days are available. Weekends aren't yours to keep, necessarily. Effectively, no "life" in work-life balance. On the other hand, Meta is missing the "work" in work-life balance. Depending on the team at Meta, desks could all be vacant by 5 pm, pre-pandemic2. Apple achieves a good balance; it doesn't worship work-life balance quite like Meta does, but burnout and mental health is still taken very seriously company-wide.
Pro #5. Exposure to leadership. This exposure is furthermore meaningful, not just via all-hands. Our team regularly presents material to leadership all the way up to the SVP, direct reports of Tim Cook's reports — in meetings of no more than 10 or so people. The leadership is furthermore engaged in these meetings.
- Impact: The team regularly checks its alignment with leadership, who also regularly give feedback on our direction and work. This both shows that leadership cares about our work and simultaneously ensures that continues to be true.
- Comparison: Unlike Tesla, regular exposure to leadership does not increase chances of being fired. Furthermore, unlike Meta, the exposure is not cursory. Meta may have weekly Q&As with the CEO, but these are large company-wide events. This awareness of what's going on, by the leadership, didn't exist even at startups. Skip levels most often never knew what I was working on.
One clear but unmentioned pro is the team itself. I absolutely love to work with my teammates — they're fun to hangout with, brilliant people to brainstorm with, and very productive, capable individuals to learn from. My manager is furthermore both technically strong and open-minded; when provided with data, he's open to proposals for even risky ideas. In short, I'm very happy with my work and team. I didn't include these as pros above only because Apple is a massive company, and your mileage may vary depending on your team.
Company culture is the main con.
The cons for Apple revolve around company-wide culture: Top-down micromanagement, internal secrecy and distrust.
Con #1. Internal secrecy. One of the company's core tenets is to "surprise and delight". This makes sense externally, but the policy also applies internally. As a result, most everyone is in the dark even about what they themselves are working on. No one knows everything, and each person holds a critical part of the puzzle.
- Even in the orientation, it's made clear that information is only shared on an as-needed basis. You don't provide more information when you think it could be helpful. You provide only if your audience has specifically requested this information and has been approved. Everyone "stays in their own lane".
- Impact: Even with a supportive manager, it becomes very difficult to connect with teams relevant to your work — in fact, it's difficult to even know which teams are relevant. If someone high up in leadership doesn't recognize the relevance, you're effectively missing out on collaborations that may significantly improve productivity. This means information is hard to come by, and the number of disclosures you have becomes a flex among product managers and external collaborators. This makes networking and corporate connections especially important within Apple, which leaders and the most senior Apple employees have reiterated repeatedly.
- Comparison: Meta's highest ratings during performance reviews is "Redefines expectations". At Apple, this is effectively impossible without any information. There are horror stories of teams asked to improve a model's performance without having access to the inputs, outputs, or a description of what the model is even doing. All they had was the ability to make code changes and rerun experiments. This means the team could not redefine the objective, problem statement, or metrics; they also couldn't clean the data or visualize qualitative performance. They were pigeonholed into improving a number, blindly.
Con #2. Top-down culture. This isn't immediately obvious, as your skip skip skip level isn't immediately dictating your everyday todos.
- Instead, the top-down culture is built by a yes-man culture. An idea proposed by an executive is immediately worshipped as divine insight; suddenly, you'll find that everyone in your chain of command is aligned on this idea. In fact, the best way to convince your manager of something is to tell them that an executive believes it.
- Impact: This means that everyone in your leadership chain is very attentive to everything their leaders say. If an executive believes it, you should too, apparently. My personal advice is to do the same, partially: Be attentive to what executives and leaders at the company are saying. You don't need to agree with everything, but executive support for an idea is a strong argument, within the company. You can use that to your advantage.
- Comparison: At Tesla, there's a similar top-down structure. However, the following is out of fear not blind admiration. At Meta, there's no such top-down structure. Many employees open criticize executives — there are many outspoken critics that I agree with but also a fair share of all-caps rude remarks.
Con #3. Recruitment is slow. In short, recruiting is not a well-documented process and team-specific. This makes the recruiting process fairly messy and slow, aggravating an already difficult job for recruiters.
- Even to transition within the company, you must re-interview as though you were an external candidate. This is a side effect of general distrust throughout the company of other teams, which we'll also discuss below.
- Impact: As a candidate for a job at Apple, this makes your offer process unnecessarily long, as we discussed in How to plan your job hunt. As a hiring manager, this also makes hiring a very drawn-out process, as you jump through hoop after hoop of approvals and documentation.
- Comparison: Naturally, the startup hiring process is much more streamlined, with better abstraction barriers between the hiring manager and the recruiters. In short, all I provided as a hiring manager was yes or no for the candidate. The process as a candidate for any other company is also much faster.
Con #4. "Apple is the best". The executives believe strongly that Apple is the best. This is certainly good in some respects — that leaders project confidence in the company. However, it borders on elitism.
- During the negotiation process as a candidate, you have to battle against the senior leader's perception that "Apple is the best. Why match or even exceed market rates?" The idea is that you should be so excited to simply be working at the best company in the world.
- Impact: This perception seeps into every team at Apple. There's both a perception that our company is the best of all companies, but there's also a perception that every team is the best of all teams. This also results in the distrust between teams.
- Comparison: Internal transfers at Meta and even Tesla Autopilot were much more fluid. There may have been an interview or two but no more than a chat with the team's manager — not a full-blown interview loop, just like for external candidates. In short, teams are more trusting of each other's performance reviews.
Join for the long haul, not for the culture.
With 20-20 hindsight, I would say that Apple is worth joining for sustainable intensity, health-promoting environment, and productivity — not necessarily for the company culture. You won't be able to learn much about Apple beyond your team, much less about how large companies work in general.
The culture is ideal for anyone ready to network or already with a large network within the company, not so much for a new graduate. Other technology companies are certainly more appealing in the short-term for new graduates, with bottom-up cultures, free food and swag, and flashier quickly-advancing products. Apple works at a slower, more regular pace.
In short, I'm enjoying my time at Apple, as it's well-suited for a long-term career. However, there are many obstacles ingrained in the company culture that make learning, growing, and thinking outside of the box difficult. If you do join, take the advice of senior leaders across the company: Learn how to network, and network like mad. You'll need your network to gather the information you need.
Why I joined Apple
Here's a brief addendum summarizing my own experiences at Apple.
In late 2021, I was looking for full-time opportunities, as my PhD was wrapping up the following spring. At the time, I had a number of compelling offers from various companies big and small — primarily in the self-driving and AR/VR spaces.
- I decided to pursue AR/VR. Quest was and still is one of my favorite entertainment consoles. With that said, I also wanted to try a new company. This narrowed it down to Microsoft or Apple.
- At the same time, given I work in computer vision, I wanted to see how one of the worlds most popular cameras and camera apps were made — either Google or Apple, then.
This made Apple a natural choice, among other factors. Now, almost a year later, I do miss parts of Tesla and Meta, but I'm still really happy with my decision overall.
Got a question? Ask me on Twitter, at @lvinwan. Want more tips? Drop your email below, and I'll keep you in the loop.