HireSight

HireSight
Team Members
Problem Statement
Solution Features
User Journey
System Architecture
Unique Value Proposition
Competitor Analysis
Competitor Analysis
Impacts
Aligned ESG
Aligned SDG
Project Timeline
Future Enhacement
Applicant's Resume Data
Conversation Log Between Interviewee and AI Interviewer, Eva
Combined Results of Interviewee's Emotional Recognition, Eye Tracking, Speech Disfluencies and Plagiarism
Interviewee's Performance Report Data
Interviewee's MBTI Detected

Inspiration

HireSight is an artificial intelligence recruiting solution designed to improve the efficiency and accuracy of talent acquisition. The platform offers a comprehensive approach to the recruitment process, leveraging advanced technologies to streamline candidate selection.

What it does

Matching Resumes:

HireSight begins by matching resumes against job requirements, using sophisticated algorithms to identify suitable candidates and their job suitability based on factors such as skills, experience, and educational background.

Natural Language Processing (NLP) Analysis:

The platform utilizes natural language processing technology to delve deeper into the content of resumes. By analyzing the applicant's skills, experience, and qualifications, HireSight provides a detailed assessment of each candidate's suitability for the role. Candidates were then ranked from highest to lowest job suitability in HR side.

AI Content and Plagiarism Detection:

HireSight incorporates AI-generated content detection and plagiarism detection tools to further verify the accuracy and authenticity of candidate resumes. This ensures that only qualified candidates with actual skills can proceed to the next stage of the recruitment process. Besides, they are also used to analyse transcripts of interview responses to identify any content that may have been copied from external sources. The aim is to ensure that the qualifications and verbal responses stated by candidates are their own and not inappropriately taken from the work of others.

Interview Process:

The interview process consists of:

AI Interview: The shortlisted candidates undergo an initial interview conducted by our AI avatar, EVA and it can dynamically change the nature of questions based on the inputs provided by applicants. During this stage, the system analyzes the candidate's responses and behavior, utilizing techniques such as emotional recognition, eye tracking such as eye movement and blink rate, as well as speech analysis such as speech disfluencies and MBTI prediction to detect potential discrepancies or dishonesty.
Technical Interview: Successful candidates from the AI interview proceed to the final round, where they are interviewed by technical personnel to assess their proficiency in relevant skills and knowledge.
Soft Skill Evaluation: Determine the candidates' soft skills during the interview to shorten the time to evaluate the talent's actual skills and characteristics.

How we built it

Research and Analysis:

Market and Technology Exploration: We conducted in-depth research into the challenges of traditional recruitment processes and understand the needs of employers and job seekers. This included identifying pain points such as hiring bias and prejudice, long hiring cycles and the high costs associated with traditional recruiting.
Technology Feasibility Study: We investigated recent advances in AI and NLP technologies to determine the feasibility of integrating such technologies into the recruitment process.

Platform Design:

We designed the architecture and functionalities of HireSight, outlining the key features such as resume matching, AI interviews, AI content and plagiarism detection, and candidate interview evaluation reports. This phase involved creating user interfaces, workflows, and algorithms to support the desired functionalities.

Model Development:

We developed various AI models to power different aspects of the platform, including resume parsing, similarity algorithms, MBTI personality prediction, facial emotion recognition, eye tracking, and plagiarism detection. We sourced high-quality data sets relevant to each model. Each model was trained and continuously fine-tuned to ensure accuracy and effectiveness. Each model was rigorously validated against a benchmark dataset to ensure they met the required performance criteria prior to integration into the platform.

Integration of Third-party Tools:

Careful selection of third-party tools is critical, especially when it comes to enhancing AI-driven conversational capabilities, which is made possible by the integration of Google Gemini in Vertex. Seamless integration of these tools through APIs is critical to maintaining smooth user interactions and data flow across the platform.

Testing and Iteration:

We conducted extensive testing to validate the performance and reliability of the platform. This involved testing individual components, as well as end-to-end workflows, to identify and address any issues or inconsistencies. We refine the platform's functionality and user interface based on the continuously feedback from our initial users.

Deployment and Optimization:

Once testing was complete, we deployed the platform, ensuring scalability, security, and performance. We continuously monitor and optimize the system to maintain its efficiency and effectiveness over time. Scalability tests were conducted to ensure the platform could handle increased loads without degradation of service.

Challenges we ran into

Data Quality:

Obtaining high-quality data for training AI models, especially for tasks like resume parsing and similarity algorithms, posed a significant challenge. Ensuring the accuracy and relevance of the data required extensive data cleaning and preprocessing efforts.

Algorithm Complexity:

Designing and implementing sophisticated algorithms for tasks such as natural language processing, facial emotion recognition, and eye tracking presented technical challenges. Balancing the complexity of the algorithms with computational efficiency and accuracy was a continuous challenge.

Integration of Third-party Tools:

Integrating third-party tools and services, such as Google Gemini in Vertex AI, into the platform required careful coordination and troubleshooting to ensure seamless functionality and compatibility with existing systems.

User Experience:

Creating a user-friendly and intuitive interface for both employers and candidates was a challenge, particularly when incorporating advanced AI features like AI interviews and facial emotion recognition. Ensuring a positive user experience while maintaining the effectiveness of the platform was a balancing act.

Ethical Considerations:

Addressing ethical considerations, such as data privacy, fairness, and bias in AI algorithms, was a crucial challenge. Ensuring that the platform complied with ethical guidelines and regulations while delivering accurate and unbiased results required careful attention and deliberation.

Scalability and Performance:

Ensuring that the platform could scale to accommodate large volumes of data and users while maintaining optimal performance was a ongoing challenge. Optimizing algorithms, infrastructure, and resource allocation was necessary to meet the demands of a growing user base.

Accomplishments that we're proud of

Efficiency Improvements:

Successfully reducing recruitment cycles and leading to faster hiring decisions and significant cost savings in human resources. Our platform frees up valuable time for recruiters by automating repetitive tasks such as resume screening and initial qualification, allowing them to focus on high-touch interactions with candidates and strategic talent acquisition initiatives.

Fairness and Standardization:

Establishing a fair and standardized evaluation process that minimizes subjective biases often found in traditional recruiting methods. Our AI algorithms assess candidates against pre-defined criteria, ensuring that everyone has the opportunity to stand out, regardless of their background or connections. This results in a more diverse and qualified talent pool, which creates a more inclusive and fairer work environment.

Enhanced Candidate Experience:

Providing candidates with a more humane and transparent hiring experience through automated tools such as AI interviews and prompt feedback collection.

Authenticity Assurance:

Ensuring the integrity and transparency of the recruitment process by implementing AI content detection and plagiarism detection tools, enhancing the authenticity of candidate information. We enable organizations to make informed hiring decisions based on a candidate's true qualifications and skills by identifying potentially fabricated information or recycled content.

Comprehensive Evaluation:

Offering a more comprehensive understanding of candidates' abilities, qualities, and adaptability through the integration of various evaluation tools, facilitating better candidate-job matching and enhancing overall job satisfaction and performance levels.

What we learned

From HireSight, we learn the significance of leveraging AI and NLP technologies to revolutionize talent acquisition processes. It emphasizes the importance of addressing challenges in traditional recruitment methods by introducing standardized tools, ensuring fairness, transparency, and efficiency in hiring. Moreover, the integration of Google Gemini in Vertex highlights the potential of advanced conversational AI models in enhancing candidate interaction and engagement during the recruitment process, paving the way for more personalized and effective candidate experiences.

What's next for HireSight

Human Interviews with AI Copilot and Analysis:

AI assistants will become integral in conducting interviews, providing real-time analysis and feedback to assess candidates' skills and adaptability accurately. When human assesses a candidate's fit with a company's culture and delves into their specific experiences, the AI assistants can analyze the candidate's responses at the same time to provide a more comprehensive evaluation.

Customized Services:

AI systems will offer tailored services based on individual needs and backgrounds, ensuring fair treatment and enhancing workplace inclusivity. This customization can extend to specific populations, such as female candidates or individuals with disabilities like blindness, further mitigating potential bias and promoting a more diverse talent pool.

Interviewee training service:

We allow interviewees to train themselves with the AI interview process before the real interview. They need to upload interview resources like the position, number of question, question type , user's resume etc. This can helps the interviewee to try and prepare before the actual one.

Lockdown Browsers:

To maintain assessment accuracy and fairness, lockdown browsers will be employed during the AI interview or while conducting online testing and assessments, preventing candidates from accessing external resources during evaluations and ensuring the authenticity of the candidate's answers.