Manupriya Gupta will be giving a practical walkthrough of evaluating AI-generated outputs, designing failure tests, and deciding when human review is needed, with examples builders can apply to their projects. For both of these sessions, participants will be learning to evaluate AI-generated outputs, design failure tests, and decide where human review belongs.
11 AM: Build a Low-Touch Travel Assistant. You call, give your location, and the assistant helps you decide where to go next and how to get there.
3 PM: Build a Voice Agent for Your Community. A more general session on building a voice assistant that helps people find community information and resources.
These are 30-minutes each, with an additional 15–30 minutes afterward for questions.
Does Your AI Actually Work? Practical Tests for Hackathon Builders
A convincing demo is a good start, but how do you know your AI works beyond the happy path? This practical session covers evaluating AI-generated outputs, designing tests that expose failures, and deciding where human review belongs. We’ll use concrete examples to help builders create a simple evaluation checklist for their own projects. Each session is 30 minutes, followed by questions. The optional starter repository for both workshops is::
https://github.com/manupriya-gupta-giga/voice-agent-starters
The repository at the above link includes the low-touch travel assistant and community voice-agent examples, with setup instructions and a few exercises. Participants can use it, adapt it, or bring their own approach. Each example can be explored independently.
For the basic demo, you only need Python 3.10 or newer and a browser. No packages or API keys are required. Browser voice features depend on support, and typed input is always available. An AI connection is optional and uses the participant’s own key.
Manupriya Gupta is a Forward Deployed Engineer at Giga (YC S23), where she brings enterprise AI agents into real-world customer workflows. She has four years of industry experience spanning algorithms, AI and quantitative finance, including 2.5 years building high-frequency trading algorithms at Goldman Sachs.
Her multidisciplinary background connects computer science, applied mathematics and financial engineering, with experience in production AI, trading systems and portfolio analytics. She holds a BTech in Computer Science and Engineering from IIT Delhi and an MS in Applied Mathematics from Carnegie Mellon University. A former teaching assistant at both institutions, she enjoys helping builders turn ambitious ideas into practical systems they can test and trust.