PROOF

is all you need.

The complete case file of Griffin Pilz — software developer.

SCROLL DOWN

[ THE SUBJECT ]

Griffin builds software that

Full-stack engineer with a data-engineering habit — pipelines, distributed systems, and AI-assisted workflows. Last seen turning hundred-billion-record days into something that doesn't flinch, then shipping the API and the UI before sunrise. Leaves the codebase cleaner than the crime scene he found.

[ 001 ] CLOSED

The 90% That Went Missing

Every compliance model started the same way: aggregating millions of enterprise emails a day — and that pull dragged on for hours. Griffin built the ingestion pipeline end to end with lexicon-based checks, deduplication, SQL Server and Oracle. Data-aggregation time for model building dropped 90%, so new models ship in a fraction of the time.

[ EVIDENCE ] Python · FastAPI · SageMaker · SQL Server · Oracle

VIEW EXHIBIT →
[ 002 ] CLOSED

Reliability, Scaling, Efficiency

Customers needed answers on tables and files — in batches and in realtime, at scale. Griffin built the inference API in FastAPI and the UI in React on top of SageMaker. It clears 100,000 inferences per second and doesn't break a sweat.

[ EVIDENCE ] FastAPI · React · SageMaker · AWS Lambda · S3

VIEW EXHIBIT →
[ 003 ] CLOSED

Meeting Compliance

GDPR and CCPA don't accept excuses. Griffin built a full-stack suppression system that found high-confidence record matches across the pipeline and pulled them in real time — no residue, no loose ends, a clean paper trail for the auditors.

[ EVIDENCE ] Go · Python · Full-Stack · Distributed Systems

VIEW EXHIBIT →
[ 004 ] ONGOING

The Agent Behind the Door

An AI agent that answers for Griffin while he's out working a case — wired into this very page. Built on the same AI-assisted workflow he uses to ship: Claude, a clear system prompt, and a healthy suspicion of hallucinations. Interrogate it from the navbar.

[ EVIDENCE ] Claude API · JavaScript · Prompt Design

INTERROGATE IT →

[ FIELD REPORT ]

[ RECORDS / DAY ]

0B+

[ INFERENCES / SEC ]

0K+

[ ETL RUNTIME CUT ]

0%

The message to everyone is a message to no one. This site is for the ones hiring.

[ OFF THE BOOKS ]

The files that never made the record.

Side projects, shipped on his own time. Every one is live — go ahead, open them up.

LIVE speedydb.org ↗

SpeedyDB

Retrieval that runs in the crawlspace of your RAM — a ~0.7 MB engine that ingests, indexes, and semantically searches documents on-device, then quietly spills to the cloud when the room fills up.

[ EVIDENCE ] On-Device · Semantic Search · RAG
LIVE referralbuddy.ai ↗

ReferralBuddy

Referrals that actually connect. Helps recruiters source, screen, and place candidates — and gets a job-seeker's resume in front of the right person, no account required.

[ EVIDENCE ] AI · Recruiting · Two-Sided
Under Investigation qa.kanposer.com ↗

Kanposer

AI ticket planning that never leaves the IDE. Break work into initiatives and phases, let Claude draft each plan, and keep execution in sync through a Cursor plugin. Still on the QA bench.

[ EVIDENCE ] Claude · Cursor Plugin · Supabase · GitHub

[ PRIOR ENGAGEMENTS ]

JAN 2026 — JUL 2026 · CONTRACT

Machine Learning & Python Data Engineer

Raymond James Financial — St. Petersburg, FL

  • Cut data-aggregation time for compliance-model building 90% with an end-to-end pipeline ingesting millions of daily enterprise emails, running lexicon-based compliance checks, and deduplicating across SQL Server and Oracle.
  • Built a SageMaker inference API (FastAPI) and React UI serving over 100,000 inferences per second — batch and realtime.
  • Scripted Python ETL moving structured data from SQL Server and Oracle into Amazon S3, with normalization and verification in Lambda for scalable model development in SageMaker.

JUL 2024 — JAN 2026

Data Engineer II

Dun & Bradstreet — Remote

  • Architected scalable data infrastructure supporting 50+ workflows and led a 6-person offshore data operations team.
  • Built custom distributed computing frameworks in Go and Python, improving large-scale validation, dedup, and transform throughput 10×.
  • Shipped a GDPR/CCPA-compliant suppression full-stack app that matched high-confidence records and removed them in real time.

OCT 2022 — JUL 2024

Data Engineer I

Dun & Bradstreet — Remote

  • Processed 100B+ records per day via normalization, mapping, bucketing, and suppression in Python, Go, and MySQL.
  • Built a Python replication system for MySQL and Redshift to sync across environments.
  • Cut ETL runtimes 50–99% by converting database ETL to multiprocessed Python.

APR 2022 — OCT 2022

Data Analyst / Engineer

American Integrity Insurance — Tampa, FL

  • Designed databases, automation, and ingestion frameworks supporting data integrity and operational efficiency.
  • Built data-collection workflows with web scraping, filtering, and transformation; automated QA and validation across multiple servers.

[ FIELD SKILLS ]

  • Python
  • Go
  • SQL
  • JavaScript
  • React
  • FastAPI
  • AWS
  • Lambda
  • Glue
  • SageMaker
  • SQL Server
  • Oracle
  • PostgreSQL
  • Redshift
  • MySQL
  • Redis
  • ETL Pipelines
  • Distributed Systems

[ ACADEMY ]

MAY 2017

B.A., Economics

University of Connecticut — Storrs, CT

[ FULL DOSSIER ]

DOWNLOAD THE FULL DOSSIER PDF · the complete record

[ THE INTERROGATION ]

Everybody has questions.

Why does he call projects "cases"?

Because every good project starts the same way: something's broken, nobody knows why, and someone has to care enough to find out. Griffin cares enough. Also, it makes the portfolio more fun.

What's he actually like to work with?

Direct, calm under pager duty, allergic to meetings that could've been a diagram. Writes things down. Ships things. Follows up.

Can he work with our stack?

The stack is a suspect, not an obstacle. He's shipped in Python, Go, and JavaScript, across SQL Server, Oracle, Postgres, Redshift, and half of AWS. He'll manage.

Is he taking new cases?

The light's on, isn't it? Leave a tip at grif@develup.com or interrogate the agent in the navbar.