💡 Our Mission

Every student has faced the daunting and nerve-wracking experience of preparing for and participating in job interviews. Particularly in the current job climate, with several large tech firms announcing major layoffs, one crucial interview can feel like it has the power to drastically shape our future. So naturally, what do we do? We put more hours into Leetcode, hoping that we can invert a binary tree during the technical interview. But this can actually work against us.

While LeetCode, AlgoExpert, and Hackerrank are great resources for future software developers to hone their programming skillset, a key differentiator for an organization when choosing a candidate is the behavioural interview. However, the tools to prepare for behavioural interviews are extremely limited or outdated. So we decided to harness GPT-3’s Natural Language Processing to build a comprehensive and intuitive tool for behavioural interview preparation. We hope our tool can help our fellow peers and hackers secure their next job offer!

🧠 What it does

InterviewPro is a powerful tool for any student that would like to improve their behavioural interview aptitudes. The user begins by providing a question they would like to work on and proceeds to provide their sample response. Next, the user will click the provide feedback button and wait for the magic to happen!

The user is taken to the analysis page. The first part of the page features a dashboard with multiple data visualizations that display various statistics, including:

  • A Radar Chart that displays the key skills demonstrated in the sample response
  • Score for the STAR Approach (Situation, Task, Action, Result)
  • Total number of filler words, and count of the most recurring ones
  • Language proficiency including clarity, coherence, and grammar.
  • The total time taken and the word count

Following the dashboard, an analysis of the key strengths and weaknesses exhibited in the response is displayed. Based on the weak areas, the user can view a personalized, first-person improvement in their response within the context of the student's original answer.

Next, the tool analyzes the top 3 skills in the response, providing feedback on what the student did well and suggesting an improvement. It also provides a specific reference to where the skill was demonstrated as well as, a reworded and improved version of that reference.

Thereafter, we provide an analysis of the candidate's approach to the STAR approach. For each component of STAR, the tool provides feedback on the strengths and weaknesses, as well as a reworded improvement of the user’s answer.

Finally, based on all the previous estimated scores, our algorithms estimate a final overall score.

🛠️ How we built it

The user provides an behavioural interview question and response. Then, the text is fed into the GPT-3 API. We use an engineered prompt designed to extract various insights and estimations through natural language processing techniques based on the datasets of interview responses and answers. Next, this data is formatted and visualized using multiple responsive react components, styled using Tailwind CSS. All API calls are done using Express.js. All images are AI-generated as well.

Tech Stack: GPT-3 API, React.js, Express.js, Node.js, TailwindCSS

🚧 Challenges we ran into

We ran into two significant problems. First, prompt engineering is a very new and emerging field, so optimizing and maximizing the output from GPT-3 API took a lot of testing. Next, connecting the backend to the frontend was complex as we haven’t previously worked with google cloud APIs.

🏆 Accomplishments that we're proud of

We are incredibly proud of this project we have built and we believe it can have a meaningful impact on our fellow peer’s career ambitions. As every student knows, acquiring experience is a daunting and stressful task, and our goal with InterviewPro was to help ease that pressure. Usually, applying for internships is not enough, you also have to ace your interview in order to pass the first "firewall". That’s where we InterviewPro comes in, we hope it can help our users build their eloquence, responsiveness and accuracy during their behavioural interviews.

📚 What we learned

The hackathon experience taught us the importance of time management, delegation, and teamwork to achieve our goals. In the short time crunch, we had to learn to trust each other, as well as, push each other to build and deliver our project. We also explored the emerging field of NLP and ML, and we are excited about the future of these technologies.

❓What's next for InterviewPro

  • Use a database to store individual users and track their progress over time. Show statistics and milestones for their improvements.
  • Incorporate Speech-to-text functionality for improved user experience and analysis.
  • Integrate an NLP tool that also improves the Resumes and Cover Letters of users, by analyzing its strengths, weaknesses, and potential improvements.
  • Develop a similar tool for technical interviews, so software interns can analyze their demos for the live coding portion of their interviews.

Built With

Share this project: