Exploring applied-AI and enablement roles · Austin, TX · remote

I ship AI tools that replace work I used to do by hand.

Eight years inside an enterprise customer-facing team at GLG, running a $14M+ portfolio. I was the AI person on a non-AI team: I built the LLM workflows, drove the adoption, then started shipping the tools myself.

Two of them are live below and you can use them right now. Every project on this page started as a problem I actually had.

See what I built Résumé ↓
Start here

Messy documents in. A read you can act on out.

samievargas.com/signal ↗

AI agent · in use · built solo

Signal

A multi-call LLM workflow that reads messy account files for second-order sentiment and surfaces revenue risk, relationship health, and pre-call briefs. I spent eight years doing that hour of digging by hand.

What you paste in

→

What comes back — 51 seconds

Try it on your own notes Anthropic API · Cloudflare Workers · source-weighted prompts
Signal output — the contact read: comm style, decision style, says vs means, how to engage

What it found — the contact read is the sleeper feature. It makes explicit what experienced managers do instinctively.

Try it here

Five minutes in. Four piles out.

samievargas.com/brain-dump ↗

Brain Dump

An AI agent that turns an unstructured dump into an energy-state-aware plan. Structured JSON out of a Cloudflare proxy to the Anthropic API, with branching logic on the state you say you are in.

The sorting is not the product. Matching the output to the state someone is actually in is the product — so change the state below and watch the same dump re-sort.

Raw dump · five minutes, no editing

Open Brain Dump
01

The analysis work

Both start from a question I actually had

The Instacart pipeline · sources through marts · 35 tests passing

Data flowing left to right
Source instacart.orders Staging · 5 models stg_order_products stg_products stg_aisles stg_departments stg_orders Intermediate int_order_products_joined fct_orders Marts dim_products dim_users

Also built

A random-forest reorder model on the same Instacart data — 0.9886 AUC for veterans against 0.8566 for new users, which is what confirmed the segment split was real and not an artifact of the pipeline. Model and notebook ↗

In progress: review-bombing detection across 31M+ Steam reviews, to catch when a score is being driven by something other than the game. The personal tools — Life OS, Life in Pixels, the tarot tracker — live on /life.

02

Experience

GLG

Austin, TX · Remote · Jul 2018 – present
8 years · 6 roles · IC to people manager
03

Skills and credentials

Plus the full Anthropic AI Fluency set — Claude 101, Claude Code 101, Agent Skills, Claude Cowork, AI Capabilities & Limitations, AI Fluency Framework — Google Data Analytics, Google Business Intelligence, GA4, Databricks Fundamentals, PMI, Six Sigma White Belt.

04

Notes from the data I live in

05

The rest of it lives on /life.

The noticing field, the tarot tracker, seven decks, the Greenbelt at fifteen of twenty-one miles, Agatha Christie in order, and what I have been listening to. Same instinct, no résumé pressure.

Go to /life →

07 · Contact

Come say howdy.

Open to anything where a messy workflow needs a system built around it.

SELECT * FROM conversations WHERE topic = 'ai'

dbt run --select samie.availability

mail sammisnv@gmail.com▍