I posted on Dev.to about the surface level stuff. The new MacBook. The part time job. The ML model I am suddenly building for a freight company when two weeks ago I could not spell XGBoost.
But I want to write the longer version here. The one where I am actually honest about what this period feels like and what I am thinking.
The Laptop First
My parents got me a MacBook Air M5.
I want to be careful about how I write this because I do not want it to sound like a flex. It is more complicated than that.
I did not ask for it. They saw me at a point and they made a call. They decided this was worth betting on. And that is a strange thing to sit with. It is not the laptop that got to me. It is the fact that someone looked at what you are doing and said yes, we think this is going somewhere.
That kind of thing puts weight on you. Not in a bad way. In the way that makes you not want to waste a single hour.
Why I Took This Job
I have been asked this a couple of times already so let me just say it clearly.
I joined a freight forwarding company as a Pricing and Tech Associate on a part time basis. And the reason I chose this over an internship is simple. Internships are where you learn. Jobs, even part time ones, are where you implement. That difference matters more than it sounds.
When you are learning, someone is absorbing the cost of your mistakes. When you are implementing, your mistakes have consequences. A client does not get their quote. A process breaks. A business loses time or money. That kind of stakes changes how you think.
I wanted that. I am at a point where I need that.
What I Am Actually Doing There
A lot. Maybe too much. But I am not complaining.
I am building a website with a CMS. I am handling LinkedIn content strategy and thinking through the growth side of the business. I am setting up n8n automations to simplify internal processes that were being done manually. And I have been tasked with building a price predictor system for freight quotes.
That last one deserves its own paragraph.
Right now the biggest operational problem is response time. A client asks what it will cost to ship something. The team has to manually work through a bunch of variables across different carriers and routes and conditions. It is slow. And in this industry, slow means lost deals.
My job is to fix that. Build something that spits out the best price fast.
I thought linear regression. Seemed reasonable. I have some data, I want to predict a number. But after actually looking at how freight pricing works I realised that was naive. Too many variables. Too much non-linearity. A simple model was not going to capture what was actually going on.
So I went down a proper rabbit hole. Used AI to help me understand the landscape. Read about different approaches. Landed on XGBoost. It handles tabular data well, works with mixed data types, does not need a huge dataset to be useful. It made sense.
I have never built an ML model before in my life. Now I am building one for a real business problem. I do not have a clean way to wrap that up. It is just true.
The Shipment Hubs
My boss has been taking me out to actually see the industry.
I have been to the Aramex hub. I am going to the FedEx hub soon. I have been learning things I never thought I would need to know. Port to port vs door to door delivery. How a shipment gets processed. Where the handoffs happen. Where things go wrong.
This is the part that I find genuinely hard to explain to people who are only thinking about this from a software perspective.
We talk a lot about building things that solve real problems. But most of the time what we mean is we sat down, thought about what might be a real problem, and built something for that imagined version of it. That is fine. That is how a lot of things get built.
But there is a different kind of thing that happens when you are actually inside an industry. You stop guessing what the problem is. You just see it. Right there in front of you. Running on a spreadsheet that it should not be running on. Taking four people to do something that one automated workflow could handle. Costing the business twenty minutes per client inquiry.
I have only been here a few weeks. I am already noticing things that could be products.
I am not going to build them all. But I am paying attention.
My Boss
I want to mention him because I think who you learn from matters a lot.
He has worked at Aramex, UPS, FedEx, Jeena, DG Group. Decades in this industry. The kind of person who has seen every version of how this space works and does not work.
He is not teaching me by handing me documents. He is teaching me by taking me places and explaining what I am looking at. That is a different kind of education. It is slower and harder to systematize but it sticks in a way that reading about something never quite does.
I am trying to make the most of being in the same room as that.
The Honest Bit About Doing Too Many Things
I know the standard advice. Go deep. Pick a lane. Specialize. Build expertise in one vertical and own it.
I have heard it. I understand the reasoning.
But I am also an economics student who taught himself to code. I have been doing case competitions. I am building a micro SaaS. I am now inside a logistics business learning how freight actually works and building ML models and running content strategy and setting up automation workflows.
None of that fits cleanly into a lane.
And I keep coming back to something I read somewhere, and I am paraphrasing badly, but the idea is that the value of being a generalist early in your career is that you find the intersections. The places where knowing two things that do not usually go together creates something neither expertise could have built alone.
I genuinely think there is something at the intersection of logistics operations and software that most people building in the SaaS space are not close enough to see. I might be wrong. But I would rather be wrong after being in the room than right from a distance.
Invoicepedia Is Just Sitting There
I have not touched it in two weeks.
I said this on Dev.to too and I want to say it again here because I think there is a version of this story where I pretend the pause was intentional or productive or part of a plan. It was not. Life got full and the project sat.
But I am not spiraling about it. The foundation is solid. I know what I want to build next. The split panel builder. The live PDF preview. The auto-save. The confetti animation on submit that I am probably more excited about than I should be.
I will get back to it.
And honestly, being inside a real business every day is going to make me a better product thinker when I do. I am watching what actual operational friction looks like. I am watching what happens when software is too slow or too manual or does not exist at all. That is going into the product instincts even when I am not sitting at my desk writing code.
What This Period Actually Feels Like
Strange. Stretched. Good.
Strange because two months ago I was purely a person trying to build things and learn things in something close to isolation. Now I am inside a company, going to shipment hubs, talking to carriers, trying to build a pricing model for a problem I had never thought about before.
Stretched because there is a lot happening across a lot of different areas and I am not going to pretend I have it all perfectly balanced. I do not.
Good because I am implementing things. Not just learning them. Not just putting them in a slide deck for a case competition and presenting recommendations to a panel of judges. Actually doing them. Shipping things that someone uses. That gap between knowing and doing is closing. That feels important.
A new laptop. A job that teaches me from the inside out. A project waiting for me to come back to it. A lot of things I do not know yet that I will know in six months.
That is where I am.
I will keep writing.