My Early Engineering Journey: Lessons from Code, Chaos, and Hackathons
The Bugs That Taught Me More Than Any Course
Every engineer has that moment — the one where a tiny bug derails your entire flow and humbles you in the best possible way. Mine didn’t come from a complex distributed system or a cutting‑edge algorithm. It came from two small, almost embarrassing issues… and a hackathon that taught me more about tools and teamwork than any course ever did.
Funny enough, these moments had nothing in common on the surface. One was about a slow PostgreSQL insert. Another was about a single misplaced % sign. And the last one? A hackathon that showed me the difference between Python and Power BI.
But together, they shaped how I think about engineering today.
When 70 Lines of Code Lost to 10
The task seemed simple: take a CSV file and insert thousands of rows into a PostgreSQL database. I reached for what I knew — psycopg2. I wrote a loop, inserted rows one by one, and felt pretty good about it.
Until I ran it.
It was slow. Painfully slow. The code was long, repetitive, and felt like something from a bygone era of Python.
Then I tried SQLAlchemy’s bulk insert.
Suddenly, 70+ lines of code shrank to around 10. The performance improved dramatically. The code became cleaner, easier to maintain, and far more elegant.
The Lesson Hidden in a Slow Script
The “familiar” tool isn’t always the best
Good engineering is often about stepping back and asking, Is there a smarter way to do this?
Abstraction, when used well, is a superpower
The Percent Sign That Broke Everything
My second challenge was even smaller — and somehow more frustrating.
I wrote a SQL query in Python using a LIKE '%1' pattern. But the query kept failing. No matter what I tried, Python refused to cooperate.
After digging deeper, I realized the issue: Python thought %1 was a formatting placeholder, not a literal string.
The fix? '%%1'
One extra %. One tiny escape character. Hours saved.
The Lesson Hidden in a Single Character
Different languages interpret symbols differently
Small characters can cause big problems
Reading error messages slowly and carefully is a skill worth developing
The Hackathon That Changed How I See Tools
A few weeks later, I joined a hackathon. I didn’t expect to learn so much about data cleaning tools. I thought “cleaning data” just meant removing a few empty rows.
Spoiler: it’s a lot more than that.
Our team had five people — me and 4 teammates. One messy dataset. One goal: turn it into a clean, meaningful dashboard.
And we approached the problem in completely different ways.
Watching Python Work Its Magic
My one other teammate opened a Jupyter Notebook and started typing. Pandas functions flew across the screen:
removing duplicates
fixing missing values
cleaning weird formatting
reshaping the data
It felt like watching someone speak a language I didn’t fully understand yet. But it was fast. Really fast.
What Python Taught Me That Day
Python is great for heavy, messy data
It gives you full control
But it requires coding confidence
Why Power BI Felt Like Home
While my teammate coded, I opened Power BI. It felt friendly — like a tool made for beginners.
With Power Query, I could:
remove blank rows
split columns
change data types
filter out unnecessary values
Everything was visual. Every transformation was traceable. It felt like building with Lego blocks.
Once the data was clean, I moved on to the fun part: building the dashboard.
What Power BI Helped Me See
Instant visual feedback
Easy undo and reorder
A smooth path from raw data to insights
Two Tools, One Team
Even though we used different tools, our work fit together perfectly.
My team member cleaned the complicated parts using Python
I cleaned the simpler parts and built the dashboard in Power BI
It felt like combining two superpowers:
Python = strong, flexible, powerful Power BI = simple, visual, beginner‑friendly
By the end of the hackathon, we had a dashboard we were proud of — and I had a much clearer understanding of when to use each tool.
Flattening the Curve of Complexity: Our Journey with COVID Survey Data
For another hackathon, we need to tackle COVID survey datasets using Python in Jupyter Notebook — three different schemas, each with its own quirks and inconsistencies. What followed was a whirlwind of data wrangling, late-night debugging, and some truly rewarding breakthroughs.
🧩 Challenge 1: Temporal Chaos
Schema 1 and Schema 2 recorded data weekly, while Schema 3 used monthly intervals. This mismatch made it nearly impossible to compare trends or analyze respondent behavior uniformly.
Our fix: We converted Schema 3’s monthly data into weekly chunks. This alignment gave us a consistent timeline across all datasets, unlocking granular trend analysis and making temporal comparisons far more insightful.
👥 Challenge 2: Age Anomalies
Age data was a mess. Schema 1 had a binary "Over_60: Y/N" flag. Schema 2 used ranges like "45–64", and Schema 3 had categorical buckets like "<26", "45–64", ">60". Aggregating or comparing age groups was a nightmare.
Our fix: We standardized age columns across all schemas, renaming them meaningfully and normalizing values into consistent age categories. This gave us a reliable foundation for age-based analysis and demographic insights.
🔍 Challenge 3: Understanding the Data Landscape
Once we cleaned and aligned the datasets, we faced a new challenge: understanding the structure and distribution of the data itself.
Our fix: We dove into exploratory data analysis (EDA). We asked questions like:
How did respondent counts vary week to week?
Which weeks had the highest participation?
How were age groups and demographics distributed?
What symptoms and travel risks correlated with probable COVID cases?
How did vulnerability differ across age categories?
Visualizations and aggregations helped us uncover patterns that weren’t obvious at first glance.
🩺 Challenge 4: Prescriptive Power
The final goal was ambitious: use the data to help doctors prioritize COVID testing when kits were limited. We needed to identify high-risk individuals based on survey responses.
Our fix: Our analysis revealed that people with pre-existing medical conditions were significantly more likely to be classified as probable COVID cases. Older adults with such conditions emerged as the highest-risk group. These insights could help doctors make informed decisions, focusing testing efforts where they’re needed most.
The Bigger Lessons Behind the Small Moments
Across all these experiences — the slow insert, the rogue %, and the hackathons — the lessons were surprisingly similar:
You don’t need to know everything
Tools are just tools
Collaboration matters
Small problems teach big lessons
Growth hides in the small stuff
Final Thoughts: Why These Moments Matter
If you’re early in your engineering or data journey, don’t underestimate the power of these small moments. They’re not setbacks — they’re stepping stones.
And if you ever get the chance to join a hackathon, do it. You’ll learn, you’ll build, and you’ll discover how different tools — and different people — can come together to create something meaningful.
Engineering isn’t built on big breakthroughs alone. It’s built on these tiny, humbling, surprisingly fun moments that push you forward.
