7 Things I Learned During My 2 Years in an AI Startup

Excitement, culture, responsibilities, and everything in between

7 Things I Learned During My 2 Years in an AI Startup

I recall the day when I was offered my first data science job. Up until then, it was mostly me learning through online courses and working on portfolio projects. On my way back home after bagging the offer, I was punching in the air and patting myself on the back for my achievement. Breaking into the world of data science as a fresher isn’t easy and is definitely an achievement. Years have passed since then. As I move into bigger challenges in my career, here’s me reflecting upon my experience and, more importantly, sharing the learnings as a Machine Learning Engineer in an AI startup for the past 2+ years.

1. Be genuinely excited and let it radiate

Read that again, with the keyword being ‘genuinely.’ Everyone claims to be excited when they join a new job. But few genuinely are and continue to be. And it makes a huge difference.

As I joined the team, I started sharing the same dream and vision of the Startup. I was fascinated by how big of an impact this small team could create. I started putting in extra effort and time to learn and improve myself because I wanted to learn more and grow faster. I was genuinely excited to complement my team and solve real-world business problems using AI. And the thought of it still excites me. I had always tried hard to control my excitement, but I would fail terribly.

During these 2 years, I started handling clients, got promoted to a Machine Learning Engineer, and finally started leading the data science team's projects. I was genuinely excited about every single opportunity that came my way, and I let it radiate. I grew, and the startup grew.

So here’s my secret, if you’re joining a startup (or any organization), be genuinely excited about the work they do, and don’t hide your excitement; let it radiate because career growth is more that way. You’ll realize this sooner or later.

2. Understand the end-to-end first

The data science life-cycle is broad. It starts from understanding the business requirements, formulating the problem, gathering and retrieving the relevant data, understanding the data, developing models to solve the problem, iterating on experiments till we optimize the results, deploying the solution, and finally communicating the results to the stakeholders. This includes different roles such as Business Analyst, Data Engineer, Data Analyst, Data Scientist, Machine Learning Engineer, and more.

I did all of this. And I'm here to tell you to do all of it too. And startups are the best place to have a go at it. Even if you don't get a chance to get your hands-on, you can always talk to your team and learn from their experiences.

In other words, become a generalist during the early stages of your career. I know I have invoked the age-old debate between generalist vs. specialist. But hear me out.

It's important to understand the end-to-end process of any job first; during this phase of your career, you will know your interests and what you're good at. Then once you know what interests you and what you’re good at, specialize in that role. In other words, start as a generalist and eventually become a specialist. This is what has worked for me all along.

3. Don’t be afraid to accept responsibilities

I didn’t do this one right. I was scared and afraid. I panicked for no reason. Now I call that a learning experience instead of a mistake.

Six months in, I was asked to lead a project end to end. It was a project on tight deadlines, and the senior management trusted me to lead it based on reasons I couldn't fathom. The thought of such responsibilities scared me to the core, and I was afraid of failing. What if the results aren't as good as they expect? What if my skills aren't good enough? Questions I asked myself purely out of fear.

As we started working on it, sure, some extra work had to be done, some roadblocks had to be broken, but we were doing well. Eventually, we branded the project as one of our own AI product offerings. The experience was one of a kind, and I have never backed down any responsibility ever since.

Expect responsibilities to come your way, earlier than you could foresee. When they come, don’t be afraid to accept them. Startups, by far, are the best place for you to learn to handle responsibilities. Always ensure to communicate your progress and any roadblocks up front, and you’ll be good to go.

4. Own your work

May it be a small feature you implement, or a dashboard you create, or a full-fledged product you build, always, always own your work.

This is true for large firms too, but very important in a startup where the teams are generally small. It’s naturally expected everyone to start taking ownership of what they do. The benefits of this are twofold, one, you learn to solve problems better, and second, it frees the supervisor to concentrate on bigger issues.

It’s often easy to raise a problem or an issue, but few follow on to solving the problem. Initially, as a junior member and later as a lead, the team found it easy to work with me when I started taking ownership of every little I do.

I keep reminding myself always, Strive to own your work always, no matter what little you do, and it will take you places.

5. Embrace the people and the culture

Irrelevant to where you work, it has to be much more than just the work. I have always believed it’s the people that are important about any organization.

When I was on a job-hunt, during my interviews, I make it a habit to evaluate their team and their culture while they evaluate my suitability for the position. Then and now, I always wanted to feel happy working where I work and with whom I work. Isn’t that what we all want?

After I joined, I embraced the people and the culture of the team. Since the team was small, without realizing much, I started bonding with everyone. Working late wasn’t hectic for me; it was fun with the right mindset. Now and then, we used to hang out and go on trips outside of work. And I still do.

I got carried away, but my point here is, please put the smallest of efforts to bond with the people and embrace the culture. It would make your life at work more enjoyable. The people I’ve worked with are the best, and I’ve earned some friends for life. I truly hope you find the same at your workplace.

6. To wear multiple hats was never a choice

I tell this to most of my mentees and on a good note. Wearing multiple hats was just inevitable. I’m not talking about the different roles within the data science life-cycle; it goes beyond that. Welcome to startup life!

So we successfully built a few AI products, but barring a few existing clients, nobody knew about it. We had to get the word out, and we didn’t have anything close to a marketing team. This is the case in most startups as they function with the core team.

Here was my line of thought. Hey, it’s our product, and we know best to market the product. We cannot trust it with some marketing firm. Remember taking ownership of your work? We ended up putting the hats of Content Writers, Marketing Executives, Web Designers, and more. We redesigned and relaunched our website with fresh content for every product to market the products we built.

Many from the team wore different hats like that, and in early-stage startups, whether you like it or not, it’s just inevitable. Learn to deal with it without losing focus on your core skill.

7. Go after excellence — quality over quantity

I’ve kept this last since this was my biggest learning of them all. When your work is of quality, in the end, trust me, it would be recognized and appreciated. The extra effort you put in would be worth it. You would evolve and become better in your craft.

It’ll be easy to deliver more in quantity when you compromise the quality of the work. It’s easy to write code that doesn’t meet code quality guidelines that does the work anyway. It’ll be easy to hack your way through all the roadblocks you face without removing them completely. You may find it easy at that moment, but you’ll soon stop feeling that way in the long run.

For example, I have made an effort to introduce software best practices, better processes within the data science team, introduce better libraries and tools to the machine learning workflow, and so on. Sure, I’d have spent some extra time on these, but these actions made my life easy when I had to maintain the products we built. So many people benefitted and thanked me in the end for introducing these amongst the team.

I can go on and on about more examples, but the bigger picture is to focus on delivering with excellence. Go after excellence. Strive to be better at what you do, showcase excellence in everything you do. In the long run, people would start to notice the quality work you produce. It’ll be all worth it, take my word for it.

Of course, this doesn’t generalize to all the AI Startups globally and is based entirely on my experience. I'm sure some of you would have different experiences too. Have you worked in an AI startup before? How different was your experience? Let’s engage in the responses section below.

What’s next for me?

I recently joined as Senior Data Scientist at a startup-like data science team in one of the largest corporates in Sri Lanka. And I can’t wait to share more of my learnings and experiences here on Medium. Also, I’ve started spending more time mentoring students to break into the world of data science.

Want to keep in touch? Join my private list of email friends.