Week 2 of my Machine Learning journey: I had to kick someone out of our group project

A hard call, and what it taught me about leading

The hard call

This week I had to remove someone from our project team. Not a technical fit with the rest of the group. We gave this person chances to catch up, but we did not have the time or capacity to both self study through the project and teach someone who could not self study on their own.

It was a difficult decision. We had invested real emotional energy into this person. But this experience gave me a perspective I had not had before.

I have only ever been an IC in my career. Never a manager. I have led pieces of a project, but never officially managed people. Eventually I will be a tech lead. I do not think I will become a manager, but I am starting to understand that at some point you have to decide to focus on the members and the work that move the project forward, and let go of the rest, for the sake of project delivery and the wellbeing of the majority of the team. I had to make that call and take full responsibility for everything that happens after.

One reason this sits heavy with me is that I was the person who, quote unquote, hired this person onto the project. So part of this is on me, on the original decision.

Another reason is that I feel responsible for keeping the project moving forward.

Some people might say, it is just a student project, why are you being so serious. I am being serious because I am putting real time into making this happen. This is going to be a real portfolio project on my resume, and I am treating every part of it seriously. How do I set up a working environment for my team. How do I show up. How do I communicate with others. These are all real practices of team dynamics and real life.

I might not be an Applied AI Engineer or Machine Learning Engineer yet, but I have worked as an engineer in a regulated industry for about five years. I am bringing those communication and people skills into this project. My team members and I can be casual with each other, we are not real coworkers, but we are professional and respectful of each other’s time and effort.

Hence, I had to let this person go.

This week’s learning

Topics: gradient descent, sentiment analysis, and feature engineering on the Reddit project.

Main resource: Andrew Ng, Intro to Machine Learning, Week 1.

Supplementary videos, organized by topic:

Sentiment analysis and NLP

Sentiment Analysis Using VADER youtube.com/watch?v=Alu_cCXNS-k

What is Sentiment Analysis? youtube.com/watch?v=5HQCNAsSO-s

Sentiment Analysis: VADER, LSTM and BERT Explained (NLP Python Guide) youtube.com/watch?v=5pqufk7kQmk

NLP Emotion Detection in Python: Compare Lexicon-based and Deep Learning Methods youtube.com/watch?v=dL7swWymrGk

Feature engineering

What is Feature Engineering | Feature Engineering Tutorial Python #1 youtube.com/watch?v=pYVScuY-GPk

Neural networks and gradient descent

But What Is a Neural Network? | Deep Learning Chapter 1 youtube.com/watch?v=aircAruvnKk

Gradient Descent, How Neural Networks Learn | Deep Learning Chapter 2 youtube.com/watch?v=IHZwWFHWa-w

This week’s building

Personal portfolio site, v1, finished, deployed, and shared with my network.

At work, I built an Oracle scripts Python application that reduced a manual process from 20 hours to 2 minutes. I also started setting up scope for a machine learning documentation matching project with stakeholders.

This week’s realization

I cannot stand hierarchy. Ever since I was a freshman in a research lab, I hated that undergrads were the bottom dog compared to PhDs and postdocs. It is one of the main reasons I never wanted to do a PhD. Big pharma has the same structure, just with more layers, many of which cannot tell the difference between an AI chat and an AI coding agent in 2026, but are very skilled at upward management.

Maybe this is the autistic in me. I do not perform politics well, and I have stopped trying to.

I only trust leaders who are either more empathetic than me or technically stronger than me, preferably both. I need a flat engineering culture. Maybe just a CTO, then several tech leads for each feature or product, and then senior and junior engineers under that. The politics and resource fighting games can be handled by TPMs or PMs, who are not directly above engineers.

This is part of why I am attracted to tech and startups. It fits who I am. I would rather work twelve hours, Monday through Saturday, remotely with technical peers I respect, than have a stable 9 to 5 where most of the job is navigating politics.

This week’s networking

I have been in a rabbit hole of ML papers, and I found several MCIT alums in this space. I reached out to them in hopes of a quick chat.

I also connected with two other programmates at MCIT. One is currently a SWE in banking aiming to obtain an MBA and work in finance. The other is an MLE in big pharma who came from wet lab experience. Both have given me new ideas and new inspiration. One taught me a new lens to do ML research, and one taught me to get plugged into the local startup scene.

Not gonna lie, I do feel salty that the Chicago startup scene is so tiny compared to SF, especially since I used to live in SF for 3.5 years.