Hi, I'm

Leonardo Apolonio

Staff ML Engineer. 10+ years building systems that work quietly at scale — NLP platforms, LLM pipelines, text analytics infrastructure. Currently at Qualtrics. Founder of Quiet Scale.

I care about the unglamorous parts of ML — evaluation, data quality, the bridge from research to production. Previously at Clarabridge, Oracle, and DARPA. MS CS from Georgia Tech.

I believe the best systems are the ones nobody notices — because they just work. Quiet Scale is the company I'm building around that idea: AI infrastructure and tools that earn trust through reliability, not hype.

Experience

Staff Machine Learning Engineer

2021 — Present

Qualtrics

Lead design and operationalization of ML/LLM text analytics platforms serving enterprise customers
Built AI-driven topic detection and conversational summarization systems
Designed LLM-as-a-Judge evaluation frameworks for production quality assessment
Analyzed ~4M customer conversations to surface systemic insights

Principal Machine Learning Engineer

2019 — 2021

Clarabridge

Modernized legacy NLP systems by migrating rules-based solutions to ML approaches
Built and deployed text classification pipelines at enterprise scale

Principal Machine Learning Engineer

2017 — 2019

Oracle

Developed ML models for natural language understanding and text analytics products

Senior Data Scientist

2016 — 2017

DARPA

Applied machine learning to defense and intelligence problems
Published research on network traffic anomaly detection using recurrent neural networks

Senior Data Scientist & Software Engineer

2013 — 2016

Advisory Board Company

Built data science capabilities from the ground up across healthcare analytics

Senior Consultant

2010 — 2013

IBM

Supported FEMA critical infrastructure and disaster recovery systems

Education

M.S. Computer Science

Georgia Institute of Technology

Machine Learning Specialization · 2016

B.S. Computer Engineering

University of Maryland, College Park

2010

Projects

Science Replication

Automated system for assessing social science research reproducibility. Analyzes papers, executes code in Docker containers, and validates results using Claude's vision capabilities.

FastAPIClaude APIDockerPython

Replication Engine

Agentic system to reproduce academic research: LangChain agent runs Python, R, and STATA via MCP executors, compares outputs to published results, and emits LaTeX reports. Supports Ollama, OpenAI, and Anthropic backends.

LangChainMCPLaTeX

PoliSignal

Live dashboard for political signal tracking and analysis — surfacing trends and metrics from public data.

Next.jsVercelData viz

Government Oversight Watchdog

Automated detection of government oversight changes from primary sources — IG vacancies, site blocks, and critical events, powered by the Wayback Machine.

Next.jsWaybackMonitoring

AI Receptionist

AI-powered phone answering for law firms — 24/7 answering, smart attorney routing, RingCentral integration, and calendar-aware availability.

Quiet ScaleLLMVoice

Quiet Scale Hedge Fund

Systematic strategy research and portfolio tooling for the Quiet Scale hedge fund initiative — infrastructure for quantitative experimentation and strategy development.

Quiet ScaleQuantResearch

Options Trading System

Automated credit-spread options on Alpaca paper: confluence across VWAP/EMA, supply & demand zones, and chart patterns; macro regime filters (VIX, sector rotation, yields); backtests, parameter experiments, daily Slack reports, and a live dashboard.

PythonAlpacaOptions