SheBoost

• 

  Project Logo

Inspiration

When starting the job search, many of us face uncertainty. We often don’t know where to begin, what tools to use, or how to effectively present ourselves to potential employers. This is especially true for women, who can encounter additional barriers in accessing job opportunities. From our own experiences and observations, we recognized the challenges in building a strong professional profile that stands out in a competitive job market.

We saw a need for a tool that could support women in enhancing their profiles, making them more competitive and increasing their chances of securing a job. SheBoost was created to help women navigate this journey by providing personalized, actionable advice to improve their resumes, LinkedIn profiles, and overall professional visibility. Our goal is to empower users with the right tools and recommendations that allow them to shine in the eyes of employers.

Additionally, traditional job platforms like LinkedIn often lack blind recruitment processes, creating inequality and bias in hiring. Platforms such as Canva may assist with design but don’t focus on improving the actual content or relevance of a professional profile. SheBoost goes beyond simple design; it offers AI-driven suggestions to help users optimize their profiles and present themselves in the best possible light.

Our work with SheBoost aligns with UN Sustainable Development Goal 5, particularly Target 5.b, which focuses on achieving gender equality and empowering women. By providing women with the tools to enhance their professional profiles, we aim to contribute to a more inclusive and equitable job market where everyone has the opportunity to succeed.

What it does

For the development of this proof of concept, we leveraged Google's cutting-edge Gemini AI technology to create an intelligent chatbot designed to offer personalized recommendations that enhance users' professional profiles. The chatbot facilitates interactions in two main ways:

1. Direct Conversation:
   Users can engage in dynamic conversations with the AI assistant to receive tailored advice about their professional development. The chatbot not only provides actionable recommendations for improving a user’s career profile but also suggests job boards and specific companies where they could apply based on their qualifications if asked by the user.
   Note: The chatbot is dedicated exclusively to career-related topics. It responds only to questions about professional growth and job opportunities. If a user asks about unrelated subjects, the chatbot will politely inform them of its specialized focus.

2. CV Analysis:
   Users can upload a PDF version of their resume, and the chatbot will analyze it to offer specific, actionable suggestions for improvement. The system also gathers relevant user data to refine its recommendations, ensuring a more personalized experience.
   Important: The chatbot is designed to process only resumes and CVs in PDF format. Any other document types will be politely rejected, with a clear explanation provided to the user.

3. The Difference from other Resume tools:
   Unlike traditional resumes created on platforms like Canva, SheBoost's approach goes beyond simple design. We focus on providing actionable, personalized feedback for users to improve their resumes and profiles in ways that make them stand out to employers.

This tool is designed to ensure users receive the most relevant advice and support in their professional journey.

How We Built It

To bring our vision to life, we followed a structured development process, utilizing a combination of cutting-edge technologies and tools. Our approach focused on collaboration, efficiency, and constant learning. Here’s a detailed look at how we developed the project:

1. Initial Brainstorming and Team Coordination
   We began with brainstorming meetings to discuss ideas, identify goals, and define the project scope. After reaching a consensus on the concept, we organized and coordinated through online meetings, as our team members are located in different regions. This approach allowed us to work cohesively despite the physical distance.

2. Design with Figma
   We used Figma to design the application's user interface, ensuring an intuitive and visually appealing layout. Figma enabled us to quickly iterate on the design, keeping it aligned with our vision for a user-friendly web application.

3. Frontend Development with FlutterFlow
   We utilized FlutterFlow to create and host the application as a web-based proof of concept. FlutterFlow’s visual development environment enabled rapid prototyping and deployment, making it ideal for this project.

4. Gemini API Integration
   To access the Gemini-1.5Flash model, we initially tested FlutterFlow's native Gemini integration for generating text and images. However, we observed that this setup limited our interactions to basic outputs, which led us to integrate the Gemini API directly via REST. This approach provided greater flexibility and control over our interactions with the AI model.

5. Prompt Optimization with Google Studio AI
   We used Google Studio AI for prompt testing and refinement, helping us configure the model to deliver more relevant and precise results. This iterative testing was crucial for fine-tuning the application’s AI responses.

6. Research on Google’s REST API
   We thoroughly studied Google’s REST API documentation to understand the request structure and how to submit documents to the model, ensuring compliance with Google’s data format requirements.

7. Secure API Key Management with Buildship
   For secure API key management, we implemented Buildship as our backend. Within Buildship, we created endpoints to manage interactions between the user and the model, allowing safe and efficient API key handling.

8. User Data Management with Supabase
   For user data storage and management, we integrated Supabase with FlutterFlow. Supabase provided a reliable backend solution that allowed for seamless data storage, retrieval, and secure handling of user data.

Challenges We Ran Into

While developing this project, we encountered several challenges that pushed us to learn and adapt quickly. These obstacles included:

1. Limited Resources with Free Accounts:
   Developing the project with free accounts for various tools and platforms created constraints, particularly when dealing with limited API usage or restricted features. This required creative problem-solving and careful planning to maximize available resources.

2. Learning New AI Tools:
   Integrating advanced AI technologies like Google's Gemini posed a steep learning curve. We had to quickly familiarize ourselves with the model’s capabilities, limitations, and how to best utilize it for our specific use case.

3. Development Time Constraints:
   With a limited development timeframe, balancing feature implementation with thorough testing and refinement was a major challenge. Time management became critical to ensure the project was delivered on schedule without compromising quality.

4. Learning Curve with Gemini and FlutterFlow:
   Both Gemini AI models and FlutterFlow were new tools for the team, and there was a significant learning curve involved. Understanding how to integrate them effectively, optimize performance, and achieve the desired user experience took time and effort.

5. Team Distance and Time Zone Differences:
   Our team members were located in different regions, which led to challenges in communication and scheduling. The time zone differences required careful coordination of online meetings to ensure everyone stayed aligned and productive.

6. Limited Backend Knowledge:
   A key challenge was the team’s limited experience with backend development. While we managed to work through this, there was a steep learning curve when dealing with API integrations, data management, and server-side configurations.

Despite these challenges, our team’s ability to adapt, learn new tools, and collaborate efficiently allowed us to successfully push through and deliver a functional and impactful solution.

Accomplishments We're Proud Of

Throughout the course of this project, we achieved several milestones that we are incredibly proud of:

1. Effective Communication Despite Geographic Distance:
   Although our team members were spread across different parts of Mexico, we maintained clear and efficient communication. Virtual collaboration played a pivotal role in ensuring we stayed aligned and focused while developing and presenting this impactful project.

2. Engagement and Learning Across Multiple Technology Domains:
   We actively engaged with and learned from a diverse set of technologies, expanding our knowledge in areas such as AI, frontend development, and backend integration. This cross-disciplinary growth was a key accomplishment for the team.

3. Leveraging Cutting-Edge Technology to Empower Women:
   We applied some of the latest and most innovative tools in the tech industry, including Gemini AI, to create a product that aims to help women improve their professional profiles and unlock more career opportunities. Contributing to the empowerment of women in the workforce is something we are particularly proud of.

4. Building a Product from the Ground Up:
   One of our greatest achievements was bringing an idea to life. From initial brainstorming to a fully functional web application, we successfully created a product from scratch, showcasing our ability to transform concepts into tangible results.

5. Stepping Out of Our Comfort Zones to Explore New Frontiers:
   Exploring the realm of generative AI was a significant leap outside our comfort zones. This project allowed us to venture into new technologies and approaches, expanding our horizons and pushing the boundaries of what we thought possible.

What We Learned

We gained valuable knowledge and developed key skills that have significantly enhanced our technical expertise:

1. Gemini Documentation and Prompt Usage:
   We learned how to effectively use the Gemini models to address specific challenges, mastering the process of crafting and refining them to ensure the best results in solving real-world problems.

2. Understanding Google’s AI Model Family: We deepened our understanding of Google's AI model family, exploring the different models available, their specific uses, and their limitations. We also learned how to integrate these models into projects, gaining a clearer insight into how to choose the right model for different tasks and optimize their integration.

3. Working with New Platforms:
   This project exposed us to a variety of new platforms, including FlutterFlow, AI Studio, Buildship, and Supabase. We developed a strong understanding of these tools and how they can be used effectively to streamline development processes and enhance project outcomes.

4. Prompt Engineering:
   A significant part of the project involved creating and refining prompts, also known as prompt engineering. We learned how to design effective prompts that elicit accurate and useful responses from AI models, a skill that is highly valuable in the field of generative AI.

5. Learning Dart Programming Language (Basic Level):
   As part of our development process, we were introduced to Dart, the language used by Flutter. We gained foundational knowledge of Dart, allowing us to integrate it into our FlutterFlow projects and understand its role in mobile and web development.

What's Next for SheBoost

As we continue to enhance SheBoost, we have several key improvements and features planned for the future:

1. Improving the User Interface Design:
   We aim to refine and enhance the user interface to ensure a more intuitive and seamless user experience. This will include visual updates and improved navigation to make the platform even more user-friendly.

2. Optimizing Interaction with Google AI:
   We plan to optimize the interaction process between the interface and Gemini AI model, ensuring faster, more accurate responses and a smoother integration that will improve the overall user experience.

3. Expanding Features for More Specialized Suggestions:
   To provide even more value to users, we intend to add features that deliver highly specialized suggestions, such as real-time job listings, acceptance of additional file types (e.g., Word documents), and the ability to generate a new document with the suggested changes. Additionally, we are considering incorporating a rating system for resumes, allowing users to receive feedback on their CVs.

4. Backend Improvements:
   We recognize the need for ongoing improvements to the backend, particularly to handle increased traffic, enhance data processing speed, and ensure scalability. We will focus on refining and optimizing the backend architecture to support future growth.

5. Transitioning to a Scalable Platform:
   To ensure SheBoost can scale effectively, we plan to rebuild the product on a more robust platform, such as React or Angular. This will provide greater flexibility, improved performance, and easier maintenance as the application grows.

6. Creating a More Powerful and Private AI Model: To increase the security and privacy of user data, we plan to create a custom, better-trained AI model using Gemini Advanced or Gemma Family model. This model will be private and exclusive to SheBoost, allowing users to share their personal and professional information without concerns about data theft or misuse. By leveraging Gemini Advanced, we will be able to further enhance the AI’s capabilities while ensuring data security, providing a safe environment for users to engage in career growth activities.

Built With

• buildship
• dart
• flutterflow
• gemini1.5-flash
• geminiapi
• google/generative-ai
• supabase