Kirthi Shanbhag

Kirthi Shanbhag

AI Engineer

San Francisco, California, United States

Actions

Kirthi Shanbhag is a AI Engineer and UC Berkeley graduate with 15+ years of software engineering experience. She's passionate about building reliable, auditable AI systems for high-stakes environments. Her recent work on TraceMind, a multi-agent NeuroSymbolic system for medical triage demonstrates how to safely combine LLMs, symbolic reasoning, and knowledge graphs. She's thrilled to be a first-time speaker at NODES 26, diving into the practices that make AI systems trustworthy.

Area of Expertise

  • Information & Communications Technology
  • Manufacturing & Industrial Materials

Safety Gates for Agentic AI: A Graph + Logic Hybrid Architecture

Most LLM applications treat the model as the decision authority.
But in high-stakes domains (healthcare, finance, ops), that's a reliability problem.

This talk walks through TraceMind, a multi-agent Neurosymbolic architecture designed for pediatric triage guidance.

It combines:
- Multi-agent orchestration (LangGraph)
- Symbolic reasoning layer (Datalog + deterministic rules)
- Retrieval-augmented generation for grounding
- Explicit safety guardrails that override model output

Key insights:
- Why you need separation of concerns (decision logic vs. reasoning),
- How RAG grounds reliability, and why explicit orchestration beats emergent behavior.
- The pattern scales beyond Healthcare, Finance, Legal, any regulated domain.
We'll show code, architecture decisions, and lessons learned shipping this system.

Kirthi Shanbhag

AI Engineer

San Francisco, California, United States

Actions

Please note that Sessionize is not responsible for the accuracy or validity of the data provided by speakers. If you suspect this profile to be fake or spam, please let us know.

Jump to top