Andre Chuabio
AI Engineer
Bringing AI to life with a focus on real-world impact, from private equity to health tech.
GitHub LinkedIn
About
As a founding AI Engineer at HelmIQ, I dive deep into building robust, AI-native CRM solutions for private markets. My work spans crafting multi-tenant infrastructure using Next.js, TypeScript, Prisma, and PostgreSQL, to meticulously managing AI spend and usage metering across Anthropic and OpenAI. I'm also on the front lines, tackling production issues to ensure smooth operations for our live clients. It's a fascinating blend of cutting-edge engineering and understanding the intricate world of private equity, a skill I honed during my time doing due diligence at EY-Parthenon.
My journey into AI engineering began at PALO IT Singapore, where I developed innovative LLM workflows on Azure, managed MCP servers, and even built a Rust transcription engine. This diverse experience laid the groundwork for my passion: health tech.
Health tech is where my heart truly sings. I've had the joy of shipping projects like MediGuard, a HIPAA-compliant DLP layer for LLMs (a hackathon winner now live on PyPI!), RehabAsCode, a unique clinician-in-the-loop rehab system, and Healthmaxx, an AI wellness coach. For me, it's less about product-market fit and more about 'outcome-market fit' – did the patient get triaged effectively? Did the protocol become safer? Did costs genuinely decrease? It's all about making a real, measurable difference.
Projects
RehabAsCode is a clinician-in-the-loop physical therapy platform featuring an AI coach, Maya, that drafts personalized protocols via a multi-agent LLM pipeline, with real-time pose estimation and wearable data integration for patient sessions.
- Clinician-approved AI-drafted rehab protocols for safety and efficacy.
- Real-time pose estimation and wearable data integration for interactive coaching.
- Deterministic multi-agent LLM pipeline for robust protocol generation and safety review.
PythonFastAPISupabasePostgreSQLVercelAnthropic ClaudeOpenAIMediaPipeTavus CVIJunction (Vital)Vanilla JSHTMLCSS
PepHouse is a health AI simulator designed to provide evidence-grounded counseling on under-evidenced medical compounds, utilizing a tiered evidence registry and a Monte Carlo twin engine to simulate patient outcomes.
- Multi-tiered evidence registry for grounding AI claims in medical data.
- Monte Carlo twin engine simulates patient outcomes with source-dependent variance.
- Tavus CVI video agent provides real-time, evidence-grounded patient consultations.
PythonFastAPISupabasePostgreSQLpgvectorReactViteTavus CVIAnthropic Claudenumpyscipy
MediGuard AI is a HIPAA-compliant middleware that safeguards patient data in LLM conversations through a multi-layered DLP pipeline, while also serving as an intelligent voice agent for conversational patient onboarding and specialist triage.
- Multi-layered DLP pipeline for HIPAA compliance, including regex, semantic, and cross-validation scans.
- Conversational AI agent for automated patient onboarding and specialist triage.
- Deployable as a standalone application or an installable MCP server for LLM clients.
PythonFastAPIStreamlitAnthropic ClaudeOpenAIBaseten (DeepSeek)You.comVoicerunVeris AI