ElevenLabs x a16z WW Hackathon

Summer Chang demos Invert View Agent, an AI interviewer built for the ElevenLabs and a16z hackathon. The tool cuts hiring costs from $4,000 per candidate by automating technical interviews with a voice agent called Mallory, then gives hiring managers a report with code, transcript, and job fit analysis.
Chapters
Hi, I'm Summer Chang.
Today I want to present to you the project we did for the hackathon event back in February 2025, hosted by ElevenLabs and x16.
So we call our project Invert View Agent. Invert means 2 streams of river meet together.
Hiring manager and the software engineer candidate, they meet together, so it's the best hiring scenario. Our teammate is Dan Guswin, Asako Haisai, and Summer Chang, me. I focus on the UI and user experience part of the project. The problem is for the software engineering hiring, it takes up 40 hours and spend company thousands of dollars, up to $4,000 for one hiring of software engineering hiring. That's huge opportunity there. We created this interview agent to help facilitate the interview process for the hiring manager to reduce the time, reduce the cost, reduce the biases, and then more precise to hire the engineer candidate for the coding challenge side of things. Those other softwares we use for the hackathon project, they are also the sponsor for the event. I'm ready to give you some live demo on the project.
As a software engineer candidate, I receive an email. I move into the next stage of interview process, which is software coding challenge. So I click to join the interview and I check the camera and I test the microphone. It sounds good. And then I consent to be recorded during the interview section and agree to process my personal data. Join the meeting. So here is how the layout looks like. I use Lovable to help me create the prompting section, which is through 2/3 of the interface, and then with 2 problems, and then time clock is counting down the time. So I can choose which problem I'm gonna solve, and I start solving the problem. Well, if I have a question, I have 11 lab Java memory to answer my question. So let's talk to Mallory.
Hi, I'm Mallory. I'll be your interviewer today for the technical software engineer coding session.
Hi, Mallory. Um, I have a question. How much time I have for the challenge?
You have 45 minutes to complete the coding challenge. Let's get started.
Okay, so it's already automatically begin when I click to enter the This interface. That's how I set it up. Now I submit the solution. It's ready. Time's up. I'm going to submit. The solution has been submitted. Then as a hiring manager, I will be receiving— oh, I'm receiving a test submission. Then it's ready to review and I will review the evaluation. Like those transcripts, the coding solution, and then the transcript from the 11 labs, and then the original problem is, and then in comparison to the job description. So I know what to expect and then calculate the results. Yeah, here's a performance review. So there's like showcase what's the skills level, and then you can explore the report. Here is helping the hiring manager to make a better decision based on what's the solution the software engineer provide.
Here is the 11th lab. I would like to share with you how I create Mallory, the chatbot. So you go to the app, and then here on the left you can see conversational AI. You create an agent, so you can simply create an agent, AI agent, here. And then mine is interview agent. Let me show you what's in here. So the language I choose English. You can choose other languages. There's up to 32 to 35 languages here you can choose. Usually I start with the first language, like first message to the candidate. Hi, I'm Mallory. I will be your interviewer today for the technical software engineer coding section. That's how the first greeting message looks like. Then I go into the system prompt. Start with overview, give an overview, and then tell the tools you're going to connect. ElevenLab 2. So there's Lovable and also, um, N-A-N is another one.
Lovable.
All right, so then you give instructions of what you want it to do. So at the end, the coding challenge is finished. Let them know and cheerfully let them know. So then I choose LLM, the fattest one. Gemini 1.5 Flash. So then you can add tools or add more knowledge-based documents here. So we're going up. You can choose the voice. So our team decided a female voice because the interview is stressful. So the female voice will be more soothing in a way, helping the candidate to complete the challenge. Coming down here, you can change this settings. You can change it as well. And then here is a widget you can copy. This is how I use to connect to Lovable. So you copy this code and then have Lovable connect to your interface. So then you can change the appearances here, change the color to match. Here's how my Mallory looks like. So yeah, you can upload an image or link or just, or, so I choose an image, a smiley Mallory. Yeah, this is my ElevenLabs interview agent setting.
Hi, so we are in Lovable interface right now and then I already know there's 3 screens I want to create. I already picked the font, picked the color, and then, so here I'm going to create, give it a little bit context, an interview agent for software technical interview. So first, to start with the first screen, receiving an email confirmation with the date, time, and link. To the next step. So for the link to the next step, need to be a button. Oh, so, and then the second screen will be to test, test the camera and the mic, and then for them to consent. Like, And the third screen will be enter the interview page with the coding on the left, the problem on the right, and the camera view on the bottom. So it's 3 screens in 1 prompt. And then, so then you can upload your color palette or otherwise they're gonna just give you a default. So you upload your color, your accent color, your typography, the animation, if you have one, and then you can refine it afterwards. I showed you earlier for the ElevenLabs Mallory, the details I created, and you can get away. And here's really simple. You just tell Lovable, here is my widget from ElevenLabs, just add it in here. So that's about it. That's Lovable.
Pretty cool. So now we are in the post. So a little bit background. information about the hackathon. It's hosted by ElevenLabs and A16Z. It's a national event hosted in 6 different countries in person, 840 participants, $92,000 in prize. Thanks to all the sponsors for the event. So if you are interested, check out the winners on Their projects. Pretty inspiring. It's really fun. And also, if you want to check out our project, you can also check out Infrared View Agent. Dan has a video. He is walking you through how the development side of things, and then our inspiration, what it does, how we build it, and the challenges we ran into. So, and also what we learned about the project.
I want to share with you a picture from the hackathon. I'm in the very corner. And then, yeah, a short clip about the whole experience for that weekend.
I'm excited to share my experience from a 25-hour in-person hackathon weekend. This was not just a marathon of coding and using AI tools. It was a deep dive into creativity, collaboration, and innovative problem-solving. As soon as the event kicked off, the energy in the room was intense. The goal was clear: push boundaries and experiment with cutting-edge AI tools. Designed and created videos in FL.AI to effectively communicate my design ideas. The event also featured presentations from the top 6 teams. Here comes the dream team to give a presentation. We didn't win or get into top 6, but it was a huge win for me.
Hey everyone, I'm excited to share that I'm partnering with ElevenLabs to create some amazing video content. If you want to support my channel and see what's possible with the tool, definitely give ElevenLabs a try. I included a link In the description box below.










