Arjhine Ty

AI Systems Engineer

I design AI systems that behave credibly under real use — multi-agent orchestration, retrieval, and the evaluation and serving layers that decide whether any of it survives contact with users.

  • clinical cases40+
  • medical specialties8
  • agents in the loop2
  • published system layers5
Selected work

Things I've built

AI systems engineering and low-level ML, shown as I actually split my time.

Sole engineerproduction

PocketPatient

A production clinical trainer where learner nurses interview AI patients that hold character across sessions: 40+ cases across 8 specialties, structured feedback, and a retrieved (not prompt-loaded) case library.

RAGNext.jsPostgres / Supabase
  • median session-turn latency1.8s
  • 30-day learner retention67%
Lead AI engineerprototype

RepReady

Enterprise roleplay where a sales rep negotiates with an AI buyer who holds a persona, objections, and history across sessions. Two agents in the loop, five published system layers, human-evaluated conversation quality.

Agent orchestrationLLM evaluation
  • conversation quality vs human roleplay4.1/5
  • median practice sessions per rep6sessions
Recognition

Awards & certificates

competition2026

ASES Manila AI Startup 101

ASES Manila

Won the ASES Manila AI Startup 101 program with Team Picaro, a clinical training pitch that became PocketPatient.

competition2025

HackHealth PH — Best Clinical AI

HackHealth PH

Placed second overall with a clinician-facing tool for medication reconciliation, built in 24 hours.

recognition2019

Deans List, College of Engineering

UP Diliman

Consistent academic recognition across four years of the CS program.

Recommendations

What people say

Arjhine is the rare engineer who treats evaluation as a first-class deliverable. The regression harness she built cut our release cycle in half, and the team still quotes her rubric standards in design reviews.

MS

Maria Santos

VP of Engineering, Northstar Health

Pairing with Arjhine on the eval-set review process changed how I think about data. She has a way of asking the question behind the question, and then measuring the answer.

JR

Jose Ramirez

Staff ML Engineer, DataSpring Labs

Her undergraduate thesis was the first time a student brought me retrieval results with the failure analysis already attached. That instinct — results and limits together — has carried into everything she has built since.

DE

Dr. Elena Cruz

Professor, University of the Philippines Diliman

Let's build something together.

Open to roles