Applying experimentation (and gambling!) to maximize your career

How you can use multi-armed bandits to find the most rewarding and fulfilling career

Welcome to all the new folks subscribing since my last post! To anyone interested in product management, career growth, and random musings consider subscribing here!

---

Time and time again people are told it is important to build up tenure in a role to be able to grow their career. When you are unsatisfied, bored, no longer learning, or whatever your workplace situation is you are told to give it time, let things settle down, everything will work out. And while this may be true over a long period of time I am here to say I believe in a completely different approach. I am telling you to be impatient, don’t wait it out, don’t give it time - be aggressive and relentless about finding what is important to you.

There is a common approach in experimentation that I think is quite relevant to how you should design your career- the multi-armed bandit. From Wikipedia the multi-armed bandit can be summarized as the following.

The name comes from imagining a gambler at a row of slot machines (sometimes known as "one-armed bandits"), who has to decide which machines to play, how many times to play each machine and in which order to play them, and whether to continue with the current machine or try a different machine.

In the problem, each machine provides a random reward from a probability distribution specific to that machine. The objective of the gambler is to maximize the sum of rewards earned through a sequence of lever pulls.[3][4] The crucial tradeoff the gambler faces at each trial is between "exploitation" of the machine that has the highest expected payoff and "exploration" to get more information about the expected payoffs of the other machines. The trade-off between exploration and exploitation is also faced in machine learning.  In early versions of the problem, the gambler begins with no initial knowledge about the machines.

The multi-armed bandit problem models an agent that simultaneously attempts to acquire new knowledge (called "exploration") and optimize their decisions based on existing knowledge (called "exploitation"). The agent attempts to balance these competing tasks in order to maximize their total value over the period of time considered.

In other words- the gambler at each bet has a tradeoff he can make- he can continue to explore and find new slot machines that are more profitable or stick with exploiting the slot machines he knows to be the most profitable so far. In the ideal solution the gambler invests some portion of his money in the existing profitable machines but continues to explore new machines with a small percent of his money (based on how much he knows about the existing machines) to find even more profitable opportunities. Early on the gambler should spend a high % of his money on exploration as he has little knowledge about each of the machines and later on as he builds up more knowledge he should spend a high % of his money on exploiting the most profitable opportunities he knows.

A similar dynamic happens with your career- early on in your career you have some signal on what you are interested in (ex. are you a math/science kind of person or a creative arts type of person, etc.). However, you have limited knowledge of what you might enjoy the best and be the best at in terms of specific roles and companies. Like the gambler early on in the game- you don’t really know what the best role for you is (even if you think you do).

Drawing from my own experience- in college I really enjoyed math / statistics / and economics. I was really good at these quantitative disciplines and thought a career that could intersect these would be the best for me. I quickly settled on trying to become a trader out of college as it sounded fun, exciting, and was aligned to my skill sets. Thus, I started my career trading soybean meal, cattle, and hog options in the pit at the Chicago Board of trade. I found this video that showcases one of the trading pits I would trade in. 

In this job I was surrounded by sweaty, angry men most of whom were not working for companies (they typically were trading their own money). I was working for the largest company and I became enemy #1 due to that mere fact. 

Everyday I would go into the trading pit and most days I would be shaking from nervousness or fear of the confrontations of the day. One thing to note, I am one of the least confrontational people you may meet and I tend to run away from confrontation. And yet my job required embracing it and fighting back on a daily basis to be able to get a part of the good trades. 

To make things worse it wasn’t just the other traders in the pit that made me nervous. My boss would also constantly be yelling at me through a headset I had to the office. He would always be yelling about one trade or another and any potential mistakes that could happen. 

All of that combined together to make me scared, nervous, and have a strong fear of making a mistake- which inevitably only caused more mistakes. I remember one trade I made where I accidentally did the math wrong in my head and took on a large position quickly. There is no time to double check things in the pit, use a calculator, or take the time to analyze the trade- it is very instinctual and primal. On this trade I immediately lost $150,000 for the company (which was double my annual salary at the time). As you can imagine my boss was enraged and yelled at me furiously.

While math / economics / statistics were all important for the role, I didn’t realize that it wasn’t a good personality fit for me when I was starting out. I did this job for 2.5 years- entirely too long. It was my first job out of college and it was one that I felt the need to try and be successful in. I was not applying the multi-armed bandit to my career - I was assuming I had perfect knowledge of the best career for me and trying to be successful despite the red flags. Looking back especially given it was my first job out of college I would have left within a year and continued my exploration. While I learned a lot in my first year, my learning substantially diminished after that so I was neither learning nor enjoying my work. However, I gained a valuable lesson from the experience - my objective function (goal) early in my career should be exploration to find what I am most excited about and learning as much as possible.

Which leads to setting the right objective function throughout your career. You can think of an objective function as what you are optimizing for- what are your goals that you want your experiments to achieve. Following are how you can think of an objective function early in your career.

Early stage (first 5-8 years out of college)

• Learn as much as possible (drives future career growth and impact)

• Discover the intersection of your passions and what you are good at- the best way to learn often is by trying new roles and seeing how you enjoy them

• Figure out what type of environment you enjoy and thrive in (small / mid / large stage, culture, consumer vs b2b etc.)

As you grow more senior and later in your career you can think more deeply about what your personal goals are- do you want to have the largest possible impact, work on interesting problems, have good work life balance etc.

How to optimize for learning

After my first role I was very clear about my objective function- I wanted to learn as much as possible and discover what role would be the best fit for me. I hypothesized I would find my niche in tech and data science – it seemed like a perfect fit for my skills and interests. Thus I landed at Salesforce as a data scientist.

At Salesforce I ended up partnering a ton with the product management team helping them analyze data to make better product decisions. I found it much more interesting than any role I had previously done and I found myself able to really thrive in the environment. I was having an impact on the product and was building great relationships with the executive team in my area.  Me taking a chance and exploring a new role was well worth it.  However, for some reason I was starting to get bored. I was better able to articulate why this was happening later on in my career and I will call this the diminishing returns of the learning curve (Image from this article):

In the beginning of any new role you are learning a ton. For me this is really exciting and you really see yourself growing. However after a while that learning curve starts to diminish and eventually flatten. What was happening to me at Salesforce was that I was hitting that point in the graph where I was still learning but it was starting to diminish - and this was causing me to become bored and restless about thinking about a new role. When you are junior in your career you generally are working on completely new experiences so you are learning a ton very quickly but after around 1 year you tend to start diminishing in your learnings - so that is the perfect time to look at new roles and companies while your objective function is still to learn. Given my new clarity of a clear objective function focused on learning and discovery of my passion I knew it was time to make a change and I ended up getting a role as a product manager at Google.

I discovered through this process that the product management role was the perfect role for me and I went on to have numerous ~1.5 year stints at various companies (Mixpanel, Lyft, and now YouTube) to hone in on the specific type of product role along with many side projects along the way. I was living the multi-armed bandit - I was learning a ton and exploring numerous different roles to figure out the best area for me. I have since started moving into more of an exploitation mindset as I have honed my skills and truly discovered the niche I want to focus on and now want to spend much longer in roles to drive large impact and exploit all my learnings.

Putting it all together

Along the way numerous mentors and interviewers asked why I would switch roles often - but I knew I was doing the best possible thing.  It is very hard to do in the moment (change is difficult!) but I have found it extremely rewarding.

Going back to the gambler at the slot machines - really evaluate how much information you have about your own skills and interests as well as how much you could gain by exploration to ensure you are making the optimal decision for your career.  You can even test this out on the side by taking on side projects / talking to people in different spaces / or doing evening classes to continue to explore while you are in your current role.  

Now switching roles frequently goes against all conventional wisdom - build tenure, stick in a role for a long time, build relationships over long periods of time in your org, and really learn a specific problem / domain.  However, I have found in my career that you can gain so much more and have so many more opportunities (as well as learn so much more) if you take the path less traveled and really explore and learn as much as possible early on in your career (and of course still build great relationships along the way).  Then once you have truly discovered your niche you can move towards longer term roles and really exploit your learnings from the diversity of your experiences towards having a large possible impact.   

So the next time you are bored in a role, feel like you are not learning a ton, or just generally feeling very comfortable - just think about whether the default you have should be to stay in your role or if there is another exciting world where you can try something completely new and continue to learn and discover your passions. Think about that gambler exploring the different slot machines trying to find the best possible one and as you explore I promise you that you will find the perfect slot machine for you.

---

Thanks for reading! And if you want more like this consider subscribing below.

Nihar Bhupalam