About Me
Applied AI Engineer specializing in audio/speech ML and multi-agent systems.
The Journey
I started in petroleum engineering in Colombia, where I built the analytical rigor that shapes how I think about systems today. Moved to New York City and pivoted to data, pursuing a Master of Science in Business Analytics at Baruch College (CUNY).
While bartending four nights a week to fund my education, I started building AI applications full-time during the day. Not demos or tutorials, but real production systems. Ten of them and counting.
The turning point was Invoz.ai. I read 46 research papers on speech processing and built a production pipeline that scores spoken English across 11 dimensions using Whisper, wav2vec2, Parselmouth, and Silero VAD. That project taught me what it means to go deep on a problem, and it became my engineering identity.
Then came Holus: a federated multi-agent orchestrator with Redis pub/sub, silo isolation, guardrails, and self-improvement loops. Building agent infrastructure taught me that the hard problems in AI are not the models but the systems around them.
Audio/Speech ML
Whisper, wav2vec2, Parselmouth, Silero VAD. Pronunciation scoring, prosody analysis, phoneme alignment. 11-dimension SQI scoring with $10.65 margin per session.
Multi-Agent Systems
LangGraph orchestration, Redis pub/sub event bus, guardrails, health preflight, observability dashboards, silo isolation, self-improvement loops.
Full-Stack & Data
TypeScript, Next.js, React, FastAPI, PostgreSQL, Redis. Real-time data pipelines, WHOOP/Strava integrations, production deployments.
Business-Aware AI
Business analytics background means I think about unit economics, pricing models, and cost-per-inference at the architecture stage, not after.
How I Work
Systems over motivation
Structures work, willpower doesn't.
Truth over comfort
Honest self-assessment, no self-deception.
One thing at a time
Depth beats breadth. Every time.
Ship then measure
Output is the proof. No performing.
Actively seeking Applied AI Engineer roles. Available immediately.