My Greatest Hits (So Far) ✨

Buckle up — these aren't just projects. They're missions, adventures, and the occasional controlled chaos that changed something real. Come see what I've been building.

01
Editorial photograph of a modern workspace with a laptop showing data visualizations.
Mission: Bias Busted 🤖

The Algorithm Whisperer

I looked at financial algorithms that were quietly closing doors on entire communities and decided — nope. This is the story of how I built a machine learning framework that hunts bias like it has a personal vendetta. Spoiler: it worked.

Read the Adventure 🔍 arrow_forwardlink
The GirlCode Chronicles 💅

Matchmaking, But Make It Tech

Young women in Africa are brilliant. Industry leaders want to mentor them. I built the platform that made the introduction — AI-matched, beautifully designed, and 100% dedicated to the idea that talent should never be held back by geography.

Live DemoGitHub Repo
02
Two diverse professional women collaborating over a tablet device in a minimalist studio.

How I Actually Build Things 🔧

No smoke, no mirrors — here's my real process from 'wait, what if...' to 'it's live and it's working and I'm screaming.'

lightbulb

Idea

First, I ask: what problem is actually worth solving? (Spoiler: it's always the human one.)

search

Research

I go deep. Like, embarrassingly deep. Data, user conversations, late-night rabbit holes — all of it.

palette

Design

This is where boring tech starts looking gorgeous. I make sure it works AND looks like it belongs on your vision board.

terminal

Building

Sleeves up, headphones in, coffee strong. Time to actually make the thing.

verified

Testing

I break it before it can break anyone else. Better me than the user.

rocket_launch

Launch

And then I set it free into the world and try not to refresh analytics every 5 minutes. (I always refresh analytics.)

THE FULL STORY

The Algorithm Whisperer: Full Story

StackPython, TensorFlow, React, AWS
Timeline6 Months (2026)

warning The Problem

The algorithm wasn't evil — it just had bad data and worse assumptions. Financial models built on historical Western data were quietly failing applicants from emerging economies, creating a loop where exclusion bred more exclusion. I couldn't let that slide.

System Architecture
Data Ingestion
Cleanse-AI
trending_flat
Core Logic
Bias Extractor
User Interface
Responsive Dashboard

Challenges

  • remove Handling sparse vector representations.
  • remove Balancing accuracy vs. interpretability.

Results

  • check_circle 34% Increase in approval fairness.
  • check_circle Reduced false-rejection rates by 12%.

Reflections

"The technical challenge is often secondary to the human one. Designing for equity means auditing our own assumptions as much as our code. This project taught me that transparency is the most powerful feature we can build."

Got Something Worth Building? 🚀

If you've got an idea that could genuinely help people, I want to hear it. Let's figure out what it takes to make it real.

Let's Make Something ✨ mail