Spaces:
Sleeping
Sleeping
File size: 12,753 Bytes
14e2173 3d7b6f6 14e2173 87a7cf5 14e2173 2f839eb 14e2173 3d7b6f6 14e2173 2f839eb e7adcda bd6074b e7adcda bd6074b e7adcda bd6074b e7adcda bd6074b 3d7b6f6 e7adcda bd6074b e7adcda bd6074b 3d7b6f6 bd6074b e7adcda bd6074b e7adcda bd6074b e7adcda bd6074b 8138173 bd6074b 8138173 2b819fe 8138173 2b819fe 8138173 2b819fe 8138173 2b819fe 8138173 2b819fe 8138173 2b819fe 8138173 2b819fe 8138173 2b819fe 8138173 2b819fe 8138173 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 |
---
title: Interview With AI
emoji: 📚
colorFrom: pink
colorTo: yellow
sdk: gradio
sdk_version: 4.27.0
app_file: app.py
pinned: true
license: apache-2.0
short_description: Mock tech interview with AI.
tags:
- LLM
- AI
- Interview
- Coding
- System Design
- Speech-to-Text
- Text-to-Speech
- Agent
- Chatbot
- Voice Assistant
- Education
---
Check out the configuration reference at https://huggingface.co/docs/hub/spaces-config-reference
# Welcome to the AI Mock Interviewer!
This tool is designed to help you practice coding interviews by simulating the real interview experience. Here you can brush your interview skills in a realistic setting, although it’s not intended to replace thorough preparations like studying algorithms or practicing coding problems.
## Key Features
- **Speech-First Interface**: Talk to the AI just like you would with a real interviewer. This makes your practice sessions feel more realistic.
- **Various AI Models**: The tool uses three types of AI models:
- **LLM (Large Language Model)**: Acts as the interviewer.
- **Speech-to-Text and Text-to-Speech Models**: These help mimic real conversations by converting spoken words to text and vice versa.
- **Model Flexibility**: The tool works with many different models, including those from OpenAI, open-source models from Hugging Face, and locally running models.
- **Streaming Mode**: The tool can use all models in streaming mode when it is supported. Instead of waiting for the full response from the AI, you can get partial responses in real-time.
## Planned Updates
This is just the first beta version, and I'm working on enhancing this tool. Planned updates include:
1. **More Interview Types**: Adding simulations like Systems Design, Machine Learning System Design, Math and Logic, Behavioral Interviews, and Theory Tests.
2. **Testing More Models**: Exploring additional open-source models to enhance the tool’s performance and flexibility.
3. **Improving the User Interface**: Making it easier to navigate and use, ensuring a better experience for all users.
# Running the AI Tech Interviewer Simulator
To get the real experience you should run the service locally and use your own API key or local model.
## Initial Setup
### Clone the Repository
First, clone the project repository to your local machine with the following commands:
```bash
git clone https://huggingface.co/spaces/IliaLarchenko/interviewer
cd interviewer
```
### Configure the Environment
Create a `.env` file from the provided Open AI example and edit it to include your OpenAI API key (learn how to get it here: https://platform.openai.com/api-keys):
```bash
cp .env.openai.example .env
nano .env # You can use any text editor
```
If you want to use any other model, follow the instructions in Models Configuration section.
### Build and Run the Docker Container
To build and start the Docker container:
```bash
docker-compose build
docker-compose up
```
The application will be accessible at `http://localhost:7860`.
### Running Locally (alternative)
Set up a Python environment and install dependencies to run the application locally:
```bash
python -m venv venv
source venv/bin/activate
pip install -r requirements.txt
python app.py
```
The application should now be accessible at `http://localhost:7860`.
# Interview Interface Overview
This tool will support different types of interviews, but currently focusing on coding interviews only. Here's how to navigate the interface:
### Setting
Configure the interview settings such as difficulty, topic, and any specific requirements. Start the interview by clicking the **"Generate a problem"** button.
### Problem Statement
The AI will present a coding problem after you initiate the session.
### Solution
This section is where the interaction happens:
- **Code Area**: On the left side, you will find a space to write your solution. You can use any programming language, although syntax highlighting is only available for Python currently.
- **Communication Area**: On the right, this area includes:
- **Chat History**: Displays the entire dialogue history, showing messages from both you and the AI interviewer.
- **Audio Record Button**: Use this button to record your responses. Press to start recording, speak your thoughts, and press stop to send your audio. Your message will be transcribed and added to the chat, along with a snapshot of your code. For code-only messages, type your code and record a brief message like "Check out my code."
Engage with the AI as you would with a real interviewer. Provide concise responses and frequent updates rather than long monologues. Your interactions, including any commentary on your code, will be recorded and the AI's responses will be read aloud and displayed in the chat. Follow the AI's instructions and respond to any follow-up questions as they arise.
Once the interview is completed, or if you decide to end it early, click the **"Finish the interview"** button.
### Feedback
Detailed feedback will be provided in this section, helping you understand your performance and areas for improvement.
# Models Configuration
This tool utilizes three types of AI models: a Large Language Model (LLM) for simulating interviews, a Speech-to-Text (STT) model for audio processing, and a Text-to-Speech (TTS) model for auditory feedback. You can configure each model separately to tailor the experience based on your preferences and available resources.
## Flexible Model Integration
You can connect various models from different sources to the tool. Whether you are using models from OpenAI, Hugging Face, or even locally hosted models, the tool is designed to be compatible with a range of APIs. Here’s how you can configure each type:
### Large Language Model (LLM)
- **OpenAI Models**: You can use models like GPT-3.5-turbo or GPT-4 provided by OpenAI. Set up is straightforward with your OpenAI API key.
- **Hugging Face Models**: Models like Meta-Llama from Hugging Face can also be integrated. Make sure your API key has appropriate permissions.
- **Local Models**: If you have the capability, you can run models locally. Ensure they are compatible with the Hugging Face API for seamless integration.
### Speech-to-Text (STT)
- **OpenAI Whisper**: Available via OpenAI, this model supports multiple languages and dialects. It is also available in an open-source version on Hugging Face, giving you the flexibility to use it either through the OpenAI API or as a locally hosted version.
- **Other OS models**: Can be used too but can require a specific wrapper to align with API requirements.
### Text-to-Speech (TTS)
- **OpenAI Models**: The "tts-1" model from OpenAI is fast and produces human-like results, making it quite convenient for this use case.
- **Other OS models**: Can be used too but can require a specific wrapper to align with API requirements. In my experience, OS models sound more robotic than OpenAI models.
## Configuration via .env File
The tool uses a `.env` file for environment configuration. Here’s a breakdown of how this works:
- **API Keys**: Whether using OpenAI, Hugging Face, or other services, your API key must be specified in the `.env` file. This key should have the necessary permissions to access the models you intend to use.
- **Model URLs and Types**: Specify the API endpoint URLs for each model and their type (e.g., `OPENAI_API` for OpenAI models, `HF_API` for Hugging Face or local APIs).
- **Model Names**: Set the specific model name, such as `gpt-3.5-turbo` or `whisper-1`, to tell the application which model to interact with.
### Example Configuration
OpenAI LLM:
```plaintext
OPENAI_API_KEY=sk-YOUR_OPENAI_API_KEY
LLM_URL=https://api.openai.com/v1
LLM_TYPE=OPENAI_API
LLM_NAME=gpt-3.5-turbo
```
Hugging face TTS:
```plaintext
HF_API_KEY=hf_YOUR_HUGGINGFACE_API_KEY
TTS_URL=https://api-inference.huggingface.co/models/facebook/mms-tts-eng
TTS_TYPE=HF_API
TTS_NAME=Facebook-mms-tts-eng
```
Local STT:
```plaintext
HF_API_KEY=None
STT_URL=http://127.0.0.1:5000/transcribe
STT_TYPE=HF_API
STT_NAME=whisper-base.en
```
You can configure each models separately. Find more examples in the `.env.example` files provided.
# Acknowledgements
The service is powered by Gradio, and the demo version is hosted on HuggingFace Spaces.
Even though the service can be used with great variety of models I want to specifically acknowledge a few of them:
- **OpenAI**: For models like GPT-3.5, GPT-4, Whisper, and TTS-1. More details on their models and usage policies can be found at [OpenAI's website](https://www.openai.com).
- **Meta**: For the Llama models, particularly the Meta-Llama-3-70B-Instruct, as well as Facebook-mms-tts-eng model. Visit [Meta AI](https://ai.facebook.com) for more information.
- **HuggingFace**: For a wide range of models and APIs that greatly enhance the flexibility of this tool. For specific details on usage, refer to [Hugging Face's documentation](https://huggingface.co).
Please ensure to review the specific documentation and follow the terms of service for each model and API you use, as this is crucial for responsible and compliant use of these technologies.
# Important Legal and Compliance Information
## Acceptance of Terms
By utilizing this project, in any form—hosted or locally run—you acknowledge and consent to the terms outlined herein. Continued use of the service following any modifications to these terms constitutes acceptance of the revised terms.
## General User Responsibilities
Users of this project are responsible for complying with all applicable laws and regulations in their jurisdiction, including data protection and privacy laws.
## Liability Disclaimer
The creator of this open source software disclaims all liability for any damages or legal issues that arise from the use of this software. Users are solely responsible for their own data and ensuring compliance with all applicable laws and regulations.
## License Compatibility
This project is released under the Apache 2.0 license. Users must ensure compatibility with this license when integrating additional software or libraries.
## Contribution Guidelines
Contributors are required to ensure that their contributions comply with this license and respect the legal rights of third parties.
## Specific Guidelines for Usage
### 1. Hosted Demo Version on Hugging Face Spaces
- **Prohibition on Personal Data Submission**: Users must not input any private, personal, sensitive information, or other restricted categories such as commercial secrets, proprietary business information, or potentially non-public financial data into this service. The functionalities that process personal data, such as CV analysis and behavioral interviews, have been disabled in this demo mode. The service is designed solely for non-personal data interaction.
- **Third-Party API Usage**: User inputs are processed using third-party APIs, including services by OpenAI and Hugging Face OS models, under the service owner's API keys. No data is stored by the service owner. Users must review and comply with the terms of service and privacy policies of these third-party services.
- **Hugging Face Policies**: Usage of this service on Hugging Face Spaces binds users to Hugging Face’s terms of use and privacy policy. Users are advised to review these policies, accessible on the Hugging Face website, to understand their rights and obligations.
### 2. Running the Service Locally
- **Absolute User Responsibility**: When the service is run locally, users have absolute control and responsibility over its operation. Users must secure their own API keys from third-party providers or opt to run local models. Users are fully responsible for ensuring that their use complies with all applicable laws and third-party policies.
- **Data Sensitivity Caution**: Users are strongly cautioned against entering sensitive, personal, or non-public information, including but not limited to trade secrets, undisclosed patents, or insider information that could potentially result in legal repercussions or breaches of confidentiality.
## AI-Generated Content Disclaimer
- **Nature of AI Content**: Content generated by this service is derived from artificial intelligence, utilizing models such as Large Language Models (LLM), Speech-to-Text (STT), Text-to-Speech (TTS), and other models. The service owner assumes no responsibility for the content generated by AI. This content is provided for informational or entertainment purposes only and should not be considered legally binding or factually accurate. AI-generated content does not constitute an agreement or acknowledge any factual statements or obligations.
|