KW KeAndre Williams Financial Analyst · Data Scientist · Systems Builder

About

About

Financial analyst by trade, systems builder by practice. I believe the best analysis is the kind someone can actually use.

Who I am

I'm a financial analyst in the Cargo division at American Airlines, working on forecasting, cost analytics, and the reporting systems that keep a global freight operation informed. I graduated from Georgia Tech with a degree in economics and a minor in Spanish. In the years since, I've built up the technical side to match the analytical one.

How I think

I believe in easy analysis.

Not simple in the sense of careless or unsophisticated, but easy in the sense that the person receiving the output can actually act on it. Complexity is not a signal of quality. A model that runs beautifully and produces a result nobody can use has failed the job, whatever the method behind it. The report that mattered is the one that answers the question clearly, on the day it's needed, in a format that fits the workflow.

Part of what shapes that instinct is that my work puts me in contact with real operations, not just data about them. I interact with people every day, visit airport stations, and watch how freight actually moves. I see how systems behave under load, and where the gap between a dashboard and the real problem tends to sit. That exposure changes how I approach what I build: once you've watched where a tool fits the workflow and where it doesn't, you start asking different questions at the design stage.

So the question I try to ask before anything ships is a simple one. Does this solve the original problem for the person who has to use it?

What I do

At American Airlines Cargo, my work spans the full analytics lifecycle, from getting the data to making it useful.

I've built machine-learning forecasting and anomaly-detection models across 10+ years of financial and operational data, covering $100M+ in annual spend. Forecasting is only useful when it's honest about its uncertainty, so I try to build things that surface problems early rather than confirm assumptions after the fact.

The data engineering side is where much of the real leverage lives. I've built automated ingestion and classification pipelines in Alteryx and Python that eliminated 100+ hours of annual manual reporting. The point isn't that automation is always better; it's that human time spent moving data by hand is human time not spent on the analysis. Clean, governed data feeds reliable models, so I treat the pipeline as part of the work rather than a precondition to it.

On the customer and commercial side, I've built segmentation and analytics models that supported the migration of 150+ enterprise customers and identified $3M+ in strategic cost savings, working directly with commercial-strategy and finance teams to turn model output into decisions.

I also work with AI and LLM tooling, integrating tools like GitHub Copilot into analytics and automation workflows. The aim is to use language models where they earn their place, not everywhere they could technically fit.

What I build outside of it

The projects on this site are the independent half, built because I wanted a specific thing to exist or wanted to understand something by building it.

One is a near-duplicate video detection tool built through three versions, each failing in a different direction. Another is a market data pipeline that generates personalized portfolio reports without anyone opening a spreadsheet. A third is a mentorship platform that handled student journals, forum channels, and credit-weighted progress tracking on Wix Velo, where I learned that the hard part of "submit a journal entry" isn't the submit button. It's the server boundary where authorship gets enforced and input gets screened.

There's also a security lab where I run attacker techniques in isolation. The goal isn't to run them; it's to see what they leave behind in the logs and write the detection that catches them.

Each project started as a question I couldn't answer without building the thing.

Certifications

  • CompTIA Security+ (SY0-701) — CompTIA. Issued April 2026, valid through April 2029. Candidate ID COMP001022971856, verifiable at verify.comptia.org.
  • CompTIA Network+ — CompTIA.
  • CompTIA CySA+ — CompTIA.
  • Alteryx Designer Core — Alteryx. Micro-credentials in General Knowledge, Data Preparation, and Data Manipulation, issued April 2025.

Georgia Institute of Technology — B.S. Economics, minor in Spanish

The résumé has the complete work history. For anything else, email is the fastest route.

At least easily, it seems.