Survival Guide for a Messy Startup Life
From a headless chicken to a small-brain chicken
Working in a big company is so easy. Pre-defined projects, mentorship, onboarding guide…while the startup is like a messy dirty muddy playground. Here are the things that I learned over the past two months in this muddy swamp.
Q: What to do first?
The biggest problem I faced, as the only person who knew about AI and machine learning, was figuring out what to do first. I also worried about whether my bosses would be happy with my choices because they don't know much about my area of expertise.
To deal with this, I came up with two solutions: a heuristic to guide my decisions, and an identity shift in how I see things.
A: 1 Heuristic - What is the laziest thing I can do to create the biggest value?
I realized that to show how valuable I am, I should focus on tasks that meet these three criteria:
Don't need too much work.
Create a big impact.
Visible enough to show the team.
I basically turned all of that into one question: “What is the laziest thing I can do to create the biggest value?” Whenever I didn't know what to do, especially when no one was giving me clear instructions, I used this heuristics to decide. In startups, it's usually more important to be fast than to make everything perfect and really big, so this heuristic considers speed more than scalability.
A: 1 Identify Shift - Think Like a Businesswoman
When I first started my job, I felt really unsettled because I thought I was doing stupid tasks (prompt engineering) that weren't technically challenging enough. But now, I prefer to think like a businesswoman and prioritize maximizing value.
Thanks to my friend who recommended a post titled "Career Suggestion Post: Don't Call Yourself A Programmer, And Other Career Advice," I began to challenge my own prejudice about "Is it just prompt engineering?" Because:
If I can ask AI to write code and replace an engineer, is it still unimportant to do prompt engineering?
If my background understanding of LLM and testing it every day allows me to do it much better than others on the team, is it JUST prompt engineering?
As they say, "Black cats, white cats, whatever can catch a mouse, it's a good cat."
Q: Am I fast enough?
"Do I work too slowly? Why do I only produce so little?"
I often wonder what "fast enough" means in a startup. Because in a startup, you always want to be faster. Then how do you define fast enough?
Additionally, I realized the reason I often can't achieve my weekly OKRs is that startups tend to have too many unexpected issues. This week, we needed to send messages to everyone and vote for our product hunt. I also had to fix new problems that arose in the production system...
Therefore, apart from my weekly OKRs, I introduced two practices:
A daily task work log to write down my daily tasks and keep track of the problems I encounter while working to be more adaptable to changes.
A time tracking system to measure my productivity.
A: Daily Work Log
My daily work log includes three pieces of information:
Daily checks (Production tasks I need to monitor every day).
Tasks for the day (What I plan to work on today).
Problems I encounter (What problems I come across and how to solve them).
With this clear record, I can easily review what I worked on yesterday and plan what I need to continue tomorrow. I find it much more helpful than a Kanban board because my tasks are often undefined and involve many small things I need to check every day.
A: Time Tracking System
Furthermore, I have also implemented a time-tracking tool to measure my productivity. For instance, this week, I accomplished quite a bit even though I didn't complete my weekly OKR. Most of my time was dedicated to monitoring production performance, resolving issues, and securing votes for our product.
This system is fantastic because all my contributions are crystal clear. For instance, yesterday, my founder said, "You haven't been proactive enough to make any changes to our AI pipeline." Instead of feeling frightened, I responded right on the spot, "That's not true. I pushed one on Monday and another one on Friday. There were no pushes in the middle of the week because we were launching our product." All of this information popped up in my brain automatically even though I was in a stressful situation.
Q: How to balance other parts of my life?
When I don't really have a work-life balance, what can I do to avoid burnout?
A: Protecting My Restful Time
When I told my friend that I struggle to balance my regular life with my hectic work life, I learned two important things:
Focus on activities that energize me the most during my downtime.
It’s a misconception to rest after I finish my work, but the opposite. Understand how much rest I need to be productive.
For principle 1 (inspired by Svitlana), it's because I have limited resting time, so instead of fulfilling all my social obligations, I prioritize recharging myself. For example, I've declined many parties I wanted to attend.
For principle 2, I've learned that in a week, I need:
Completely off-work time on the weekend.
Exercise 3-4 times a week.
Engage in at least one in-person and one online life-sharing session with friends (focused on learning, growth, and discussing challenges).
Spend time writing about what I've learned.
Reflect on my week.
These practices are necessary for me to function well. So rather than thinking, "I'll rest when I finish my work," I now plan for a certain amount of rest and specific activities to ensure I stay productive.
Q: What if I'm wrong and lost?
Two mindsets that I've learned to embrace when feeling lost and fearing mistakes are the experimental mindset and the leadership mindset.
A: Experimental Mindset - It's Okay to Be Wrong
Startup life is incredibly chaotic and confusing because things change so rapidly every day. Typically, startups want their employees to be fast and competent, but the best part about startups is that they are very forgiving of mistakes. It's often better to ask for forgiveness than to seek permission.
Since I've always been afraid of making mistakes, I made an agreement with my founders: I'll take action and maybe make mistakes initially, but they should inform me when something is not okay. This way, I can overcome my perfectionism and stop guessing what my founders need.
A: Leadership Mindset - Trust Me, I Can Do It
In a startup, I must take the lead in the AI and ML direction because I'm the only one with expertise in these areas. However, I often struggle with self-doubt, especially when I tell myself that I'm a recent graduate, an intern, and lack sufficient AI experience and knowledge. This self-doubt often prevents me from leading our product or trusting my judgment.
My friend, Ha, posed two questions to me:
Can your founder or anyone else on the team do your job better than you can?
Yes, initially, but not anymore.
Can they easily find someone else to do your job?
No, it's difficult to find a replacement due to the challenges of finding talent, my deep understanding of the system, my alignment with the team's values, and my relationships within the team.
So, I realize that I'm valuable. I am the founding engineer for our AI, and even though I'm new, I am capable enough to lead and make decisions for our AI/ML direction.
Closing thoughts
As the only person handling the core of our product, I was constantly freaked out, scared, lost, or confused. I thought I was a young new grad, I thought I was an intern, and my founder thought I couldn’t adapt well to this messy environment.
But I started to learn that those doubts were not true after I started to develop all these heuristics, systems, and mindsets (I started to fight with my founder and I don’t care haha).
Yesterday when my founder asked in our 1:1, “Do you think the startup environments don’t have enough structure for you to strive?” I immediately replied without hesitation, “No, I took the initiative in X, Y, and Z last week already.”
I’m here to be the best AI engineer and to build great things. I am the core of our product and I’m irreplaceable. Even though my founder controlled my return offer, but screw it. I will not even let him question my ability.
Because I’m not a headless chicken! I’m a chicken with a small brain!
Note: Thanks to Dev, Phuong, Ha, and Brandon, for giving me advice in this confusing process.
AI is so racist it thinks you're white bcz you sound smart af in this post =))