ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β A LIFE IN MISMATCHED PASTELS β
β Every color misaligns just enough to make meaning β
β Looks accidentalβuntil you see the pattern β
ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
FDF2F8 ββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
FFE0F5 ββββββββ 2022: Lab coat meets unexpected pastels βββββββββββ
D4FFE4 ββββββββ Where pink met mint met purpose ββββββββββββββββββββ
Name: Cazzy A.
Current Role: Head of Data @ FoXX Health
Background: Quality Control Scientist β Ethical AI
Trajectory: Former lab scientist who traded pipettes for Python
Education:
- MS Data Science (University of Denver)
- BS Integrative Biology & Chemistry (OSU Cascades)
- AI in Healthcare Certificate (Johns Hopkins, 2025)
Mission: Building AI that addresses healthcare inequities for women
Specialty: Pattern discovery in distribution tails & bias detection
Philosophy: Every model must be validated, evidence-based & production-ready
Approach: Effectiveness + Attractiveness + Impact = Excellence
# My ideal palette: Mismatched pastels that shouldn't work but doI started in a lab coat, where I learned that good science means obsessing over validation and reproducibility. Turns out, those habits translate pretty well to machine learning. Iβm here to make sure we are building ethical AI. In womenβs health, βgood enoughβ models still fail real people, so my work is bias audits, subgroup calibration, and ruthless validation...and then shipping tools people actually use. I like the weird edges of data: tails, drift, the places fairness breaks. Iβm stubborn about ethics and practical about delivery. Iβll trade a headline metric for a safer model every time and, Iβll show you why with evidence, not vibes. Why me? I bridge research and production. I write the tests, instrument the monitors, and say βnoβ when the data canβt support the claim. Bring me the messy dataset youβve been avoiding; Iβll tell you what the tails are saying, and weβll make it useful together. If you care less about hype and more about calibration curves, weβll get along. I like turning messy data into useful, fair systemsβmodels that explain themselves, pass their audits, and still look good in a dashboard. If youβre curious about outliers, tail behavior, and pushing code that doesnβt quietly exclude half the population, say hi. |
D4FFE4 ββββββββββββββββββββββββββββββββββββββββββββββββββββββ
93C5FD ββββββββ 2019: Data Science emerges βββββββββββββββββ
FFCCE5 ββββββββ 30% reduction in errors ββββββββββββββββββββ
|
Where I learned that |
Discovered Python doesn't |
Promoted twice |
Started fixing bias in |
Now designing |
A7F3D0 ββββββββββββββββββββββββββββββββββββββββββββββββββββββ
FFCCE5 ββββββββ 2024: Lead Data Scientist ββββββββββββββββββ
93C5FD ββββββββ Built ML platform from scratch ββββββββββββββ
def career_acceleration():
"""
The gradient mismatches intentionally.
Knowledge compounds in unexpected colors.
"""
timeline = {
"2024_Q1": "Lead Data Scientist",
"2024_Q2": "Architected frameworks",
"2024_Q3": "AI Engineer",
"2025": "Head of Data",
"gradient": "exponential",
"palette": "Always mismatched"
}
return "Where patterns emerge from chaos" |
%%{init: {'theme': 'base'}}%%
graph TD
A[Dataset] -->|r = 0.06| B[Everyone: No Pattern]
A -->|But...| C[Me: Check the Extremes]
C -->|Top 5%| D[r = 0.85!]
C -->|Bottom 5%| E[r = 0.85!]
D --> F[PATTERN HIDDEN IN EXTREMES]
E --> F
F --> G[Being evaluated for drug safety & financial risk]
style A fill:#FFE0F5,stroke:#D4FFE4
style C fill:#D4FFE4,stroke:#93C5FD
style D fill:#93C5FD,stroke:#FFCCE5
style F fill:#A7F3D0,stroke:#E6E0FF
style G fill:#34D399,stroke:#FFE0F5
|
|
Bridging the gap between "Hello World" and production code. |
An AI assistant that adapts its thinking style. |
Infrastructure that composes itself. |
The stuff that actually ships to production |
Where math meets aesthetics |
Because models need homes too |
Because learning new languages keeps me curious |
|
From measuring chemical reactions to measuring algorithmic bias. |
THE GRADIENT OF HARM:
βββββββββββββββββββββββββββ Clinical trials exclude women
ββββββββββββββββββββββββββ 8/10 drugs affect women differently
ββββββββββββββββββββββββββ 50% higher misdiagnosis rate
ββββββββββββββββββββββββββββ Real people harmed daily
MY INTERVENTION (IN MISMATCHED PASTELS):
ββββββββββββββββββββββββββββ Detect bias (pink on mint)
ββββββββββββββββββββββββββ Balance data (blue on blush)
ββββββββββββββββββββββββββ Fair models (lavender on sage)
βββββββββββββββββββββββββββ Healthcare for all (in every shade)
|
| I'm always interested in conversations about pattern discovery, ethical AI, or why medical algorithms think everyone is a 70kg male. Also happy to discuss career transitions, the beauty of well-documented code, or why pastel color schemes are objectively superior. Hidden patterns in data β’ Building fair AI systems β’ Healthcare innovation |
βββββββββββββββββββββββββββββββββ Curious about patterns?
ββββββββββββββββββββββββββββββββ Interested in fairness?
ββββββββββββββββββββββββββββββββ Want to build together?
ββββββββββββββββββββββββββββββββββββ Let's make AI fair
|
|
|
|
|
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
β β
β COMPRESSION COMPLETE: β
β β
β Career = β«(Pink β Blue β Mint β Purpose)dt β
β β
β Every color combination intentionally unexpected β
β Pink on mint, blue on blush, lavender on sage β
β Mismatched but never unintentional β
β β
β I find patterns in noise β
β I fix bias in algorithms β
β I do it all in mismatched pastels β
β β
β Because different is powerful β
β And unexpected is memorable β
β β
βββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββββ
%%{init: {'theme': 'base', 'themeVariables': {'primaryColor':'#FFE0F5','fontSize':'14px'}}}%%
pie title Where My Code Lives
"Python (Data Science)" : 45
"Python (ML/AI)" : 30
"JavaScript (Viz)" : 15
"R (Stats)" : 8
"Shell (Automation)" : 2
I write code like I used to write lab reports: obsessively documented, thoroughly tested, and with enough comments that future-me won't hate past-me. |