MockTalk.ai

This is what our website homepage would look like!

Inspiration

Our inspiration was going through technical interviews ourselves and realizing that information and problems requires to prep for these interviews is often hard to find and practice with efficiently.

What it does

We created MockTalk.ai in order to make interview information(such as Leetcode problems and behavioral interview tips) more accessible to the user and to help them receive smart feedback on their answers based on our AI model. It takes in the user's interview date, dream role and company, and what experience level they wish to apply as to give solid feedback to the user so that they can be well prepared for the big day.

How we built it

We built this using OpenAI's GPT-3.5 Turbo model, as well as Langchain as our LLM to train our model. The backend of this code was written using Python, and we are using Flask and HTML, CSS, and JavaScript for our frontend, as we plan to host this on a website.

Challenges we ran into

Some challenges we ran into include figuring out how to interact and code with the OpenAI API, as it took some time to understand how the agent interpret's user responses. We also ran into issues with Langchain, as our memory was not working and it took time to understand how to work with the OpenAI agent.

Accomplishments that we're proud of

We are especially proud with having a prototype-working backend using Python, as it holds and maintains smart conversations with users. We also are really proud of coming into this hackathon without any AI/LLM working experience and learning the ins and outs of how AI APIs work, how to create a fullstack project, and what tools/frameworks are needed to build it.

What we learned

We learned a lot about how to work with AI APIs and LLMs, as well as how to work together with a large fullstack project and distribute action items with each other. It was also interesting learning about tokens and how AI models really work like.

What's next for MockTalk.ai

MockTalk.ai aims to expand from just giving questions and feedbacks from technical interviews to other roles in other industries. We also look forward to solidifying our website and integrating our API and LLM with the frontend completely so that we have a completed product.

Built With

css
flask
github
html
langchain
openai
python
serpapi

Submitted to

UC Berkeley AI Hackathon

Created by

I worked on a little backend and frontend, and worked with integrating the OpenAI API model with the rest of our Python code, as well as learning Flask and HTML/CSS to stylize our site. I led the brainstorming process of what features we wanted to include with our project, and learned more about utilizing technologies I've never used before, such as Langchain and APIs. Though we did not finish, I learned a great amount of AI knowledge and how to use LLMs/APIs to create something cool, as well as learn from other amazing hackers.

Aditi Chatterjee
I worked on the backend. It was my first time using Python since my experience has been in C++ and a little bit of Java. I made the contributions in the LangChain_Implementation github branch, which translated the base python code into LLM notations. I did not finish, and it was my first time using LangChain and anything with AI. This is also my first hackathon, but I did love the experience and knowledge I gained from this. I also helped Beatrice with debugging and giving ideas for the base code. It was also my first time using Github!

Anna Doan
I came up with the idea of creating an interview bot and worked primarily on the backend of the code and figured out the usages of ChatGPT with Anna. I also took a look at some front end aspects with Aditi and Charlize. Although I'm most comfortable with Java, it was my first time truly working with AI and Python as well as LLM's. Although we were not able to finish our code completely, I'm looking forward to going back on our code and further working on it!

Beatrice Aragones
I worked on a little of the backend at the start where we tried to call the OpenAI API calls as i had done previous API projects. Further on we tried to learn how to do the front-end with flask as a micro-framework, as none of us had substantial experience in building a web application! Though we did not finish, we had a much better understanding of what it takes to create a web application

Charlize Tan

Updates

Aditi Chatterjee started this project — Jun 18, 2023 05:28 PM EDT

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.