Tiantian Zhan

Building agentic AI infrastructure for voice, healthcare, and human-agent interaction.

AI SYSTEMS BUILDER
AGENTIC AI · VOICE · HEALTHCARE
she / her
tiantian [dot] zhan [at] yale.edu
Tiantian Zhan

WHAT I'M SHIPPING

  • Production Voice Agent Runtime

    ASR · LLM · TTS pipeline deployed to ~200K smart devices.

  • AIOS / Agentic Infrastructure

    Capability-centric routing, agent memory, multi-LLM orchestration.

  • Human-Agent Interaction Research

    Multi-agent intervention for academic procrastination.

About

Yale MPH in Health Policy with an applied-math background — focused on AI systems and healthcare AI.

Currently building production voice AI and agentic infrastructure for smart devices and enterprise AI applications.

Long-term threads: agentic AI, human-agent interaction, healthcare AI, voice AI, and AI systems in production.

Selected Work

Five projects that capture the range — from production infrastructure to HCI research.

  • AIOS
    AGENTIC INFRASTRUCTURE

    Unified capability layer for agentic AI — multi-model orchestration, agent memory, MCP-native runtime.

  • Real-time Voice Agent Runtime
    VOICE AI · HARDWARE

    Production ASR–LLM–TTS pipeline with edge-cloud co-inference. Rust + WebSocket, ~200K wearables in the field.

  • HCI-Agent
    HCI RESEARCH

    Multi-agent intervention for academic procrastination — Big Five persona, social norms, and responsibility pressure.

  • MedRAG-LLM
    HEALTHCARE AI

    Clinical knowledge–augmented framework. RAG over medical literature for clinical-report analysis.

  • LLaVA-Med / Biomedical VLM
    MULTIMODAL

    Vision–language alignment for medical imaging — chest X-ray analysis and radiology reasoning.

Research

Ongoing threads, working papers, and research notes.

  • Human-Agent Interaction

    Multi-agent behavioral intervention, procrastination, social influence.

  • Healthcare AI

    Clinical reasoning, biomedical NLP, RAG over medical knowledge.

  • Multimodal AI

    Biomedical vision-language models, multimodal content analysis.

  • Efficient Adaptation

    LoRA / PEFT / prompt tuning, adapter-based fine-tuning.

Writing

Notes on agentic AI systems, voice AI for hardware, healthcare AI, and AI in production.

First posts coming soon.

CV

Full résumé as PDF — quick contact below.