Project Highlights (internal)
We have developed most features of their base product and more recently developed a GenAI-based candidate evaluator called HiredNow. You can see a demo in the below link. The idea is to give the AI agent a character (for example, an angry customer at a bank), and see how a potential candidate (for this example, considering the position of a bank teller) interacts with the character. To do this we built a bespoke simulation engine with LLMs to guide the conversation and eliminate hallucinations of the AI agent (in many ways we effectively built our own langchain but specifically for simulations). They are actually going to open-source a portion of the LLM simulation engine we built.
Link - https://drive.google.com/file/d/1zFaBiYCASi0LMMF-Jkpdqen7X7tFoufb/view?usp=sharing
This is a digital survey product and they wanted to see if we could use LLMs to predict answers to future surveys based on past survey responses from users. We developed a platform called Synthetic that does this and provides scores for new survey answers based on thousands of previous user responses. Demo in the link.
Link - https://drive.google.com/file/d/16KdPyr3xtwpj0KuONqF_cidnav4FKJPo/view?usp=sharing
We have developed an AI chat extension to this product that can answer any user question about the data that is available on the platform. The demo in the below link shows how it generates answers based on the data that is crowd-sourced in the IdeaScale platform used by Tesla.
Full-stack capabilities (Trymata and Skillfully)
These are two examples of pure web app building that we have done that are in production rights now and used by hundreds of users.