"Beyond the Prototype: Building Resilient, Non-Hallucinating AI Agents at Google Scale"
"Beyond the Prototype: Building Resilient, Non-Hallucinating AI Agents at Google Scale"
Wed May 13 - 6:50 PM EDT/New York (See in local time)
Add to Calendar
05/13/2026 6:50 PM
05/13/2026 07:10 PM
America/New_York
#WTGC2026
"Beyond the Prototype: Building Resilient, Non-Hallucinating AI Agents at Google Scale"
#WTGC2026
"Beyond the Prototype: Building Resilient, Non-Hallucinating AI Agents at Google Scale"
https://www.womentech.net/ringcentral
https://www.womentech.net/ringcentral
"Beyond the Prototype: Building Resilient, Non-Hallucinating AI Agents at Google Scale"
"Beyond the Prototype: Building Resilient, Non-Hallucinating AI Agents at Google Scale"
Don’t miss out and join visionaries, innovators, and thought leaders from all over the world at the Women in Tech Global Conference.
Vote by Sharing
Unite 100 000 Women in Tech to Drive Change with Purpose and Impact.
Do you want to see this session? Help increase the sharing count and the session visibility. Sessions with +10 votes will be available to career ticket holders.
Please note that it might take some time until your share & vote is reflected.
Session: Beyond the Prototype: Building Resilient, Non-Hallucinating AI Agents at Google Scale
Building an AI agent prototype is often just 5% of the journey. The challenge lies in the remaining 95%, transforming that prototype into a resilient, inclusive, production-ready system capable of serving millions of users without hallucinating.From the real-world experiences of building agents, this session explores the engineering required to bridge the prototype to production gap. We will dive into the critical pillars of production-grade AI: building robust evaluation frameworks to tame non-determinism, implementing multi-layered safety (redactions, citations, and consent) and infrastructure guardrails to protect diverse users from bias and harm, defining hard metrics for reliability, and establishing monitoring and debugging when models drifts. Attendees will learn to architect agents that are factually grounded, inclusive, and reliable at scale using industry-standard patterns applicable to any tech stack.
Key Takeaways
- Adopt a Production-First AI Mindset: Shift your engineering focus from raw model capabilities to the critical infrastructure: evaluations, safety, and observability, required to scale agents safely and inclusively.
- Build Automated Evals and Guardrails: Deploy multi-dimensional auto-raters to measure relevancy, faithfulness and fairness, and implement strict safety guardrails to prevent hallucinations and data leaks to protect diverse users.
- Define AI-Specific SLIs/SLOs and Cost Controls: Establish actionable metrics for latency and error rates, and implement protective engineering (token quotas, rate limiting, AI credit systems) to prevent runaway LLM costs.
- Architect AI Observability and Incident Response: Instrument your agents to trace thought trajectories, detect model drift, and build emergency stop and hot-patching mechanisms to mitigate live generative AI outages.
Bio
Susmita Singh:I am a Senior Software Engineer at Google leading efforts in AI infrastructure for the Google Store. With a focus on building evaluation frameworks, safety guardrails, and scalable orchestration, I am dedicated to building resilient generative AI agent experiences that are factually grounded and reliable for millions of users worldwide. I am deeply passionate about mentoring women in engineering and hope to use the platform to demystify complex AI systems for the next generation of technical leaders.
Anamika Modi:
I am a Senior Software Engineer on the Google One team, where I lead the integration of Google’s flagship AI models into the consumer subscription ecosystem. I led the development of Google’s AI Credit System and orchestrated the onboarding of multiple AI partners, including Antigravity, Video Generation tools, Notebooklm. My work focuses on the intersection of AI capabilities, subscription monetization, and production-grade monitoring for high-scale AI features. Outside of my core technical work, I am an active advocate for inclusive tech communities and am thrilled to share actionable, open-source-friendly architectures with the GHC audience.
