BondsAI | Devpost

BondsAI Lobby
Chat Interface for the Applicant
Assessments/Summary received by the recruiter on the Recruiter Dashboard

Inspiration

BondsAI didn't start off as an AI screening APP. At first, we thought of integrating AI into a dating app, in a way that allows the AI to analyse a person's traits from a simple conversation. However, realised that a bigger, more useful application for this mechanic is in corporate hiring. As AI continues to improve rapidly, even job applicants have started using AI to apply for jobs. They use AI to adapt their resumes to different jobs, creating more opportunities for them, but simultaneously, this vast volume of applications for each role puts immense pressure on each company's hiring team. Furthermore, current manual screening and interviews are time-consuming, inconsistent, and often subject to human bias—especially in highly competitive fields. Our product, BondsAI, aims to stand on the hirer's side and fix this problem. We will fight AI with AI, dwindling down this vast volume of submissions whilst maintaining fairness for all applicants and hirers, as well as improving the interview experience and efficiency.

What it does

Through our chat interface, the job candidate has a one-on-one discussion with our AI chatbot. Through the candidate's responses to the AI's specific questions, the AI gains a better understanding of the candidate's personality and qualities, measuring specific attributes ranging from their teamwork ability, to their experience to their technical skills, to their cultural fit and adaptability. At the end of the interview, the candidate is presented with a confirmation screen and the information is sent to the recruiter interface. The recruiter can then access the information through their interface by clicking on each applicant's bubble: inside are the scores of all the different metrics, which lead to the total score, as well as the justification for the candidate's score.

How we built it

BondsAI is a modular, async Python application for AI-powered job screening and assessment. The backend leverages Flask for API endpoints, OpenAI’s async client for scalable AI interactions, and environment-based configuration for secure deployment. Core business logic is encapsulated in data-driven classes, with robust parsing, scoring, and report generation. The frontend is served as static HTML/CSS/JS, providing real-time candidate and recruiter interfaces. Utilities handle session management, file I/O, and timing. The architecture emphasises separation of concerns, extensibility, and efficient, non-blocking workflows—delivering a seamless, automated interview and analytics platform for quantitative trading recruitment.

We used Cursor and Cline to make the rough base structure of our project, then manually debugged, improved and refined the project (both the frontend and backend).

Challenges we ran into

The biggest challenge was seamlessly integrating the backend and the frontend, making sure that the AI's messages would be displayed on the screen. Another challenge we ran into was getting the code to parse the assessment file: while the text file generated would have the correct numbers in it, these numbers would not be accurately displayed on the recruiter frontend (they would show a score of 0).

Accomplishments that we're proud of

We're proud to have created a product that can give such a detailed and insightful overview of a candidate based on a brief conversation the candidate had with the AI chatbot. We also think that our project is complemented by the UI design (the logos, the colours, the movement of the different elements).

What we learned

The Power of Async and Modularity: Leveraging async programming and modular design made our system scalable, maintainable, and responsive—key for real-world AI applications.
AI Integration Challenges: Integrating with large language models is powerful but requires careful prompt engineering, error handling, and validation to ensure reliable, relevant results.
Importance of Data Structure: Using data-centric classes and clear separation of concerns helped us manage complex candidate data and business logic cleanly.
User Experience Matters: Both candidate and recruiter interfaces need to be intuitive and robust; even the best AI needs a smooth user journey to deliver value.
Configuration and Security: Centralising configuration and using environment variables is essential for secure, flexible deployment—especially when handling sensitive data and API keys.
Iterative Development: Building a product like this is an iterative process—testing, refining, and learning from each step, especially as we adapt to real user feedback and edge cases.
Ethics and Fairness: Automating hiring brings responsibility; we learned the importance of transparency, fairness, and bias mitigation in AI-driven assessments.

What's next for BondsAI

Our current plan aims to fully integrate BondsAI screening into our university clubs' recruitment processes first. As a safe and controlled space for testing and implementation, we can gather real data and feedback to improve our product before moving into the industry.

How we imagine BondsAI in 10 Years:

Industry Standard: BondsAI could become a standard platform for automated, unbiased, and data-driven hiring, trusted by top firms globally.
Multimodal AI: The system may leverage advanced multimodal AI, analysing not just text but also voice, video, and behavioural cues for richer candidate profiling.
Personalised Assessments: Adaptive, hyper-personalised interviews that adjust in real time to each candidate’s responses, background, and even emotional state.
Global Reach: Supporting multiple languages and cultural contexts, enabling fair and effective hiring worldwide.
Integration Ecosystem: Deep integration with HR, ATS, and professional networking platforms, providing seamless end-to-end hiring workflows.
Continuous Learning: The platform could use ongoing feedback and hiring outcomes to continually refine its models, ensuring ever-improving accuracy and fairness.
Ethical AI Leadership: BondsAI could help set the bar for ethical, explainable AI in hiring, with transparent scoring and bias mitigation.
Beyond Hiring: Expanding into employee development, team formation, and organisational analytics, using AI to optimise not just hiring, but long-term talent success.

In short, BondsAI could evolve from a smart screening tool into a comprehensive, AI-powered talent intelligence platform, shaping the future of work and recruitment.