portfolio · 2026
Aaron Gayle
Data + AI curious. Builder. Occasional photographer. A decade in marketing taught me to listen to data; now I let it drive.
about
A bit about me
Computer-science trained, then a decade in Tokyo running campaigns where the most interesting part was always the data — performance analysis, bid optimisation, trend hunting in noisy logs. These days that's the whole job: analysis, machine learning, and shipping things that make the numbers honest.
On the side: hardware tinkering, portrait photography, and the occasional long run.
work
Where I've been
-
2022 — Present
Customer Experience Manager · teads
Performance Analytics & Campaign Data
- In-depth performance data analysis to identify trends and lift campaign effectiveness
- Data-driven post-campaign reports and presentations for top-tier clients
- Quarterly strategy informed by campaign performance data across DSPs
-
2018 — 2022
Sales Manager · Info Cubic Japan
Revenue Reporting & Market Analysis
- Sales tracking, reporting, and revenue forecasting to leadership
- Emerging-market identification through competitive and trend analysis
- Multi-million-dollar budget management against quarterly KPIs
-
2017
Team Leader · Info Cubic Japan
Team Performance Reporting
- Comprehensive team-performance reporting and KPI tracking
- Cross-team coordination on project QA and operational improvements
-
2015 — 2016
Digital Marketing Consultant · Info Cubic Japan
Bid Optimisation & Big-Data Trend Analysis
- Exploratory data analysis to drive campaign performance improvements
- Processing big data to identify trends across PPC/CPM channels
- Bid management, keyword research, and ROI/ROAS optimisation
-
2011 — 2015
BSc Computer Science · Goldsmiths, University of London
Foundations: Programming, Algorithms, Data
- Programming foundations in Java, Python, SQL, HTML/CSS
- Statistical and computational thinking underpinning later data work
projects
Selected work
- Sample AUC 0.89
Customer Churn Predictor
Placeholder — an end-to-end gradient-boosted churn model deployed behind a lightweight API.
- Python
- XGBoost
- FastAPI
- MLflow
- Sample IoU 0.76
Satellite Land-Use Segmentation
Placeholder — semantic segmentation of satellite imagery for land-use classification.
- Python
- PyTorch
- Segmentation
- GeoSpatial
- Sample Live App
Daily Habit Tracker
Placeholder — offline-first web app for tracking daily habits with gentle streaks.
- TypeScript
- React
- IndexedDB
- PWA
- Sample Open Source
Markdown Notes CLI
Placeholder — a zero-config terminal tool for personal markdown note-taking with full-text search.
- Go
- CLI
- SQLite
- FTS5
- Sample Top 5%
Ames Housing — Top 5% Finish
Placeholder — Kaggle regression challenge finishing in the top 5% of submissions.
- Python
- LightGBM
- Stacking
- Feature Engineering
- Sample F1 0.84
NLP Disaster Tweets
Placeholder — distilled transformer for real-vs-fake disaster tweet classification.
- Python
- Transformers
- DistilBERT
- NLP
photography
Through the lens
Photography is the slower side of what I do — a quieter site, optimised for bigger images. The full set lives there.
Visit Photography Site →-
Desert light
-
Nocturne
-
Linen
-
Carbon
-
Gold hour