Chapter I · December 2023 · AI intern, working solo
Can a language model judge how well a call‑center agent handled a call?
The MVP
I built it on my own.
Record the call. Transcribe it. Diarize it, so every line has a speaker. Evaluate it with an LLM.
Fully automated QA.
- Record01
- Transcribe02
- Diarize03
- Evaluate04
Record. An 8-NIC capture server decodes Adtran UDP/SIP streams and G.711/G.729 audio into stereo WAV.
Transcribe. Speech recognition runs on NVIDIA NeMo.
Diarize. NeMo and pyannote.audio split the call into agent and caller.
Evaluate. An LLM reads the diarized transcript and evaluates how the agent handled the call.
Chapter II · May 2024 – January 2025
AI InternAI Engineer
The only AI engineer at AnSer.
In production
Built alongside
A dashboard and a router.
A Next.js dashboard that gives supervisors and agents visibility into calls.
An emergency-call routing system built with Twilio, Python and Azure PostgreSQL. It generates revenue.
May 2024
B.S. Computer Science.
University of Wisconsin–Green Bay.
Chapter III · January 2025 – now
AI EngineerLead AI Engineer
No longer the only AI engineer. The team is growing.
I own the AI stack at AnSer, architect the infrastructure for every new product and lead cross-functional teams.
September 2025 · My own product
Season Approved
A personal color analysis app that helps people shop their season.
What it does
It scores retailer products against each user’s seasonal palette.
Where it runs
Native iOS in SwiftUI. Native Android in Kotlin/Compose. And the web.
Chapter IV · 2026
How I work now
Scoped tasks. Parallel sessions.
I split a project into scoped tasks and run several AI coding sessions in parallel.
Every change passes CI, tests and code review before it merges.