‭Assessment for AI Agent Project‬

‭Internship Details:‬
‭Stipend‬‭: €200-400 per month, based on experience and performance‬
‭Duration‬‭: 3 months‬
‭Location‬‭: Remote‬

‭Project Deadline and Submission Details‬

‭‬ D
● ‭ eadline‬‭: November 18th 2024‬
‭●‬ ‭Submission‬‭: Please send your completed project, including the GitHub repository link, to‬
‭[email protected]‬

‭ elcome to the assessment! This document provides a complete overview of the requirements‬
W
‭and expected results, along with answers to potential questions. This project will help us evaluate‬
‭your skills in machine learning, API integration, and prompt engineering, as well as your approach‬
‭to creating user-friendly applications.‬

‭Project Overview‬
‭ ou will create an AI agent that reads through a dataset (CSV or Google Sheets) and performs a‬
Y
‭web search to retrieve specific information for each entity in a chosen column. The AI will leverage‬
‭an LLM to parse web results based on the user's query and format the extracted data in a‬
‭structured output. The project includes building a simple dashboard where users can upload a file,‬
‭define search queries, and view/download the results.‬

‭Expected Deliverables‬
‭1.‬ G
‭ ithub Repository‬‭: A well-organized repo containing the project code, organized into‬
readme.md‬‭file that explains the setup, usage, and key features‬
‭relevant directories, with a‬‭
‭of the project.‬
‭2.‬ R
‭ eadme File‬‭: This should include a project summary, setup instructions, usage guide, and‬
‭details on any third-party APIs or tools you used.‬
‭3.‬ L
‭ oom Video‬‭: A 2-minute video walkthrough explaining the project’s purpose,‬
‭demonstrating key functionalities, and highlighting any notable code implementations or‬
‭decisions.‬
‭4.‬ ‭Working Dashboard‬‭: A simple, intuitive UI that allows users to:‬
‭○‬ ‭Upload CSV files or connect a Google Sheet.‬
‭○‬ ‭Choose the primary column with the list of entities (e.g., companies).‬
‭○‬ ‭Input a custom prompt for information retrieval.‬
‭○‬ ‭View and download the extracted results.‬

‭Breakout Consultancy Private Limited‬

‭CO: Neue Rothofstraße 13-19, 60313 - Frankfurt am Main, Germany | CIN: U70200UT2024PTC017145 | E: [email protected] |‬
‭www.breakoutinvesting.in | Tel: +49 151 6858 3271‬
 ‭Assessment for AI Agent Project‬

‭Key Features & Requirements‬

‭ ere’s a breakdown of each feature you’ll be building, along with details on the expected outcome‬
H
‭for each:‬
‭1. Dashboard for File Upload and Google Sheets Connection‬
‭●‬ ‭Goal‬‭: Allow users to upload a CSV file or connect to a Google Sheet directly for data input.‬
‭●‬ ‭Expected Outcome‬‭:‬
‭○‬ U
‭ sers can upload a CSV file with a simple "Browse" button or enter Google Sheet‬
‭credentials to link to a sheet.‬
‭○‬ D
‭ isplay the available columns from the CSV or Google Sheet, allowing the user to‬
‭select the main column (e.g., company names).‬
‭○‬ ‭Show a preview of the uploaded data.‬
‭●‬ ‭Technical Details‬‭:‬
‭○‬ I‭ntegrate with the Google Sheets API for real-time Google Sheet access, allowing‬
‭users to authenticate and pull data from their sheets.‬
‭2. Dynamic Query Input with Prompt Template‬
‭●‬ G
‭ oal‬‭: Allow users to specify the type of information they want to retrieve for each entity in‬
‭the main column.‬
‭●‬ ‭Expected Outcome‬‭:‬
‭○‬ A
‭ text input box where users can define a custom prompt, such as "Get me the‬
‭email address of {company}".‬
‭○‬ E
‭ nsure users understand they can use placeholders (e.g.,‬‭{company}‬
‭) that will be‬
‭dynamically replaced by each entity in the selected column.‬
‭3. Automated Web Search for Information Retrieval‬
‭●‬ G
‭ oal‬‭: Perform web searches for each entity using the custom prompt and gather relevant‬
‭web results.‬
‭●‬ ‭Expected Outcome‬‭:‬
‭○‬ F
‭ or each entity in the selected column, your agent should conduct a web search‬
‭(e.g., “Get me the email address of {company}”).‬
‭○‬ ‭Gather and store search results (e.g., URLs, snippets) for each entity.‬
‭○‬ C SerpAPI‬
‭ onsider using‬‭ ScraperAPI‬
‭,‬‭ ‭, or other web scraping/search APIs,‬
‭ensuring proper API usage limits.‬
‭●‬ ‭Technical Details‬‭:‬
‭○‬ ‭Implement logic to handle rate limiting and avoid potential blocking.‬

‭Breakout Consultancy Private Limited‬

‭CO: Neue Rothofstraße 13-19, 60313 - Frankfurt am Main, Germany | CIN: U70200UT2024PTC017145 | E: [email protected] |‬
‭www.breakoutinvesting.in | Tel: +49 151 6858 3271‬
 ‭Assessment for AI Agent Project‬

‭○‬ S
‭ tore each entity's results in a structured format, ready for further processing by the‬
‭LLM.‬
‭4. Passing Results to an LLM for Parsing and Information Extraction‬
‭●‬ G
‭ oal‬‭: Use an LLM to extract specific information based on the user-defined prompt and‬
‭web results.‬
‭●‬ ‭Expected Outcome‬‭:‬
‭○‬ S
‭ end each entity’s search results to the LLM, along with a backend prompt like‬
‭“Extract the email address of {company} from the following web results.”. This‬
‭prompt could be asked from the user as well.‬
‭○‬ E
‭ nsure the LLM processes the search results and extracts the requested‬
‭information (e.g., email, address, etc.) for each entity.‬
‭●‬ ‭Technical Details‬‭:‬
‭○‬ I‭mplement LLM integration (e.g., Groq or OpenAI’s GPT API) for processing the‬
‭data.‬
‭○‬ H
‭ andle any errors gracefully, such as retrying failed queries or notifying the user if‬
‭data retrieval is unsuccessful.‬
‭5. Displaying and Storing the Extracted Information‬
‭●‬ G
‭ oal‬‭: Show extracted data in a user-friendly format and provide an option to download the‬
‭data.‬
‭●‬ ‭Expected Outcome‬‭:‬
‭○‬ D
‭ isplay the extracted data in a table format within the dashboard, organized by‬
‭entity and extracted information.‬
‭○‬ O
‭ ffer an option to download the results as a CSV or update a connected Google‬
‭Sheet with the extracted information.‬
‭●‬ ‭Technical Details‬‭:‬
‭○‬ ‭Provide a “Download CSV” button and an option to update the Google Sheet.‬

‭Optional Advanced Features‬

‭For extra credit, consider implementing the following:‬
‭●‬ A
‭ dvanced Query Templates‬‭: Allow users to extract multiple fields in a single prompt, such‬
‭as “Get the email and address for {company}.”‬
‭●‬ G
‭ oogle Sheets Output Integration‬‭: Offer users the option to write back the extracted data‬
‭directly to the Google Sheet.‬
‭●‬ E
‭ rror Handling‬‭: Robust error-handling mechanisms for failed API calls or unsuccessful‬
‭LLM queries, with user notifications.‬

‭Breakout Consultancy Private Limited‬

‭CO: Neue Rothofstraße 13-19, 60313 - Frankfurt am Main, Germany | CIN: U70200UT2024PTC017145 | E: [email protected] |‬
‭www.breakoutinvesting.in | Tel: +49 151 6858 3271‬
 ‭Assessment for AI Agent Project‬

‭Technical Stack Suggestions‬

‭ elow is a list of recommended technologies, but feel free to use alternatives if you’re comfortable‬
B
‭with them:‬
‭●‬ ‭Dashboard/UI‬‭: Streamlit, Flask‬
pandas‬‭for CSV files; Google Sheets API for Sheets‬
‭●‬ ‭Data Handling‬‭:‬‭
‭●‬ ‭Search API‬‭: SerpAPI, ScraperAPI, or another search service‬
‭●‬ ‭LLM API‬‭: Groq or OpenAI’s GPT API‬
‭●‬ ‭Backend‬‭: Python‬
‭●‬ ‭Agents:‬‭Langchain‬

‭Submission Instructions‬
‭1.‬ G
‭ ithub Repository‬‭: Push your project to a public GitHub repository. Make sure to organize‬
‭your code and include a comprehensive‬‭ README.md‬ ‭.‬

readme.md‬‭should contain:‬
‭2.‬ ‭Readme File‬‭: Your‬‭
‭○‬ ‭Project Description‬‭: Brief overview of the purpose and capabilities of your project.‬
‭○‬ ‭Setup Instructions‬‭: How to install dependencies and run the application.‬
‭○‬ U
‭ sage Guide‬‭: Instructions for using the dashboard, including connecting Google‬
‭Sheets and setting up search queries.‬
‭○‬ A
‭ PI Keys and Environment Variables‬‭: Explain where users should enter their API‬
‭keys and any other required environment variables.‬
‭○‬ ‭Optional Features‬‭: Highlight any extra features you’ve added.‬
‭3.‬ ‭Loom Video‬‭: Record a 2-minute video walkthrough of your project, explaining:‬
‭○‬ ‭The overall purpose of the project.‬
‭○‬ ‭Key features and how the dashboard works.‬
‭○‬ ‭Any notable code implementations or challenges you encountered.‬
README.md‬
‭○‬ ‭Share this video link in your repository’s‬‭ ‭.‬

‭Breakout Consultancy Private Limited‬

‭CO: Neue Rothofstraße 13-19, 60313 - Frankfurt am Main, Germany | CIN: U70200UT2024PTC017145 | E: [email protected] |‬
‭www.breakoutinvesting.in | Tel: +49 151 6858 3271‬
 ‭Assessment for AI Agent Project‬

‭Frequently Asked Questions‬

‭1. Can I use any other search or scraping API?‬

‭ es, you may use any search or scraping API of your choice as long as it retrieves web data‬
Y
‭effectively and can handle automated requests. Some popular options include‬‭ SerpAPI‬‭,‬
‭craperAPI‬
S ‭, or even search APIs provided by major search engines if they allow sufficient‬
‭requests for testing.‬

‭2. What if I can’t find a free LLM API?‬

I‭f you’re unable to find a free or accessible LLM API for testing, you can use the‬‭Groq API‬‭, which‬
‭offers a free tier specifically for projects and experimentation. We encourage you to use this option‬
README.md‬‭for how users can‬
‭if you’re unable to access other LLMs. Include instructions in your‬‭
‭set up a Groq API key if you choose this option.‬

‭3. What if the LLM fails to extract the needed information?‬

I‭mplement a fallback mechanism, such as retrying the query or notifying the user that the‬
‭extraction was incomplete. Consider displaying a message in the dashboard to indicate that data‬
‭may be missing for some entities and make sure errors are handled gracefully.‬

‭4. Are there restrictions on UI design?‬

‭ eep the UI simple, clear, and user-friendly. While we suggest using‬‭Streamlit‬‭for a quick setup,‬
K
‭Flask‬‭is also acceptable if you prefer more control over UI customization. The focus is on‬
‭functionality, so prioritize features and usability over aesthetic design.‬

‭5. How should I handle API keys and other sensitive information?‬

‭ tore sensitive information like API keys in environment variables for security purposes. In your‬
S
README.md‬
‭ ‭, provide instructions on where users should input their API keys and any other‬
‭required credentials. Avoid hardcoding these details directly in your code.‬

‭6. What if I need more time or run into technical difficulties?‬

‭ lease reach out to us if you encounter specific technical issues or if you require more time to‬
P
‭complete the project. We're here to support you, so feel free to communicate any blockers.‬

‭7. Are there limits on what data I should scrape or extract?‬

‭ ocus on retrieving publicly available information related to each entity (e.g., email, address, or‬
F
‭products). Respect website terms of service, and do not attempt to scrape sensitive or private‬
‭data. Always use API services responsibly, adhering to their usage guidelines and rate limits.‬

‭8. How do I handle large datasets efficiently?‬

‭Breakout Consultancy Private Limited‬

‭CO: Neue Rothofstraße 13-19, 60313 - Frankfurt am Main, Germany | CIN: U70200UT2024PTC017145 | E: [email protected] |‬
‭www.breakoutinvesting.in | Tel: +49 151 6858 3271‬
 ‭Assessment for AI Agent Project‬

I‭f processing large datasets, consider limiting the number of queries or implementing a batch‬
‭processing approach. This will help avoid timeouts and API rate limits while making your solution‬
‭scalable.‬

‭9. Can I add extra features or enhancements?‬

‭ bsolutely! We encourage you to be creative with optional features (e.g., handling multiple‬
A
‭prompts in a single query, more advanced error handling, or exporting results to Google Sheets).‬
README.md‬
‭Just make sure to clearly highlight any additional functionality in your‬‭ ‭.‬

‭Stay Connected‬‭:‬
‭Follow‬‭Kapil Mittal‬‭on LinkedIn for updates and opportunities:‬‭LinkedIn Profile‬‭.‬
‭Join the Telegram channel for updates in trading and investment:‬‭Telegram @breakoutinvesting‬‭.‬
‭Check out other trading insights and strategies on TradingView:‬‭TradingView Profile‬‭.‬
‭Visit the Breakout AI website for more about what we do:‬‭breakoutai.tech‬‭.‬

‭Breakout Consultancy Private Limited‬

‭CO: Neue Rothofstraße 13-19, 60313 - Frankfurt am Main, Germany | CIN: U70200UT2024PTC017145 | E: [email protected] |‬
‭www.breakoutinvesting.in | Tel: +49 151 6858 3271‬
 ‭Assessment for AI Agent Project‬

‭Breakout Consultancy Private Limited‬

‭CO: Neue Rothofstraße 13-19, 60313 - Frankfurt am Main, Germany | CIN: U70200UT2024PTC017145 | E: [email protected] |‬
‭www.breakoutinvesting.in | Tel: +49 151 6858 3271‬