The work
Proof, not promises.
Every project, with the data behind it.
The new-business boom, and its caveat
Twenty years of Census data: applications are up big since 2019 — but a shrinking share look like future employers.
+58.8%
more business applications per week vs 2019
Reprocessing only what changed
An incremental Parquet lake that refreshes one partition instead of rebuilding everything.
15%
of the lake touched per refresh — not the whole history
The weekly report that writes itself
freddyxai's own metrics report, regenerated every Monday on a cron — tracker included, $0/month.
16 s
weekly cloud run · $0/mo
One month of NYC 311, cleaned and proven
A 331,976-row public dataset made analysis-ready, with the validation report to prove it.
100%
rows passing all rules (was 99.99%)
A $0 warehouse, run by GitHub Actions
4,090,836 taxi trips modeled, tested, and re-run monthly — for $0/month.
$0/mo
to run (4,090,836 rows)
Deduping records, scored against the truth
Fuzzy-matched a messy restaurant dataset — and proved it with a held-out test split.
0.9333
F1 on held-out test data
One dashboard, a million receipts
A 1,067,371-row retail dataset analyzed and shipped as a live single-view KPI dashboard.
96.8%
of revenue from repeat buyers
freddyxai SEO/GEO baseline
Established the search + citation baseline for a new agency domain.
0→Page 2
target keyword
Cited by ChatGPT, on purpose
The play to get freddyxai named in AI answers — running now.
0 of 12
citations at baseline
What a public dataset says about SMB churn
A finding the ICP cares about, pulled from open data.
1
public dataset
One workflow, automated end to end
Replaced a manual weekly report with an agent.
5 hrs
saved weekly
A $0 data pipeline a non-engineer can run
A small, free pipeline anyone on the team can operate.
$0
monthly cost
The newsletter
Receipts in your inbox.
Every build and post, as it ships. No fluff.