Christian Vega · Senior Data Analyst · Data & AI Engineer

I build data systems end to end.
Three of them are live.

I design, build and operate data systems end to end — the architecture, the pipelines, the application, and the operations that keep them running. Three of them are mine and in production: LicitaPR, StreetOS and Nidopr. Six years in regulated healthcare, payer and government data, currently as a Senior Data Analyst at Optum (UnitedHealth Group).

0 live production systems
0 listings served (Nidopr)
0 contracts ingested (LicitaPR)
0 healthcare & enterprise data

The proof, in numbers

What these systems do.

~64,000 property listings aggregated & served (10+ sources, nightly) Nidopr
+44% → +5% valuation-model bias after back-testing & trust-gating Nidopr ML
40 automated production health checks → phone / Discord / email Nidopr Monitor
~1.37M government contracts ingested from 34 sources LicitaPR
422K → ~345K vendors resolved to entities (Splink + deterministic) LicitaPR
Multi-tenant food-business OS: ordering, KDS, inventory, tax, close StreetOS

See all the work →

How I work

One accountable owner,
from raw data to a running product.

Six years across healthcare, payer and government data, and three production systems taken from the first crawl to live users. I design the architecture, build the pipelines and the models, ship the application and run the operations that keep it alive — and I hold every claim to a test before it ships. That is what a client gets from Vizlogic, and what I bring to a team.

Off the clock

Can you see a breakup coming in the aggregate?

A question I couldn’t shake, chased down with 16 years of real US search data and hand-rolled statistics — then I spent most of the effort trying to prove my own answer wrong. It even cites the peer-reviewed paper that argues against its own thesis.

16 yrsof real US Google Trends, Jan 2010 – Jul 2026 (~9,900 monthly records)
4 / 4bucket next-month (t+1) forecasts that beat seasonal-naive AND survive Benjamini-Hochberg FDR (July 2026 refresh)
0.354strongest t+1 out-of-sample skill (reset, Diebold-Mariano p < 0.0001); distress 0.284 at p = 0.0036
See what I built for fun →
databro.vizlogic.tech View live ↗
Move the month, watch the gauge settle, run the forecast out, then read the caveats sitting right next to the charts.

Have data? Let’s make it think.

Available for consulting engagements through Vizlogic, and open to senior data, analytics and AI engineering roles — remote, U.S.