Embedded Files
Problem
Problem
For job seekers in learning design, adding "AI proficiency" to a resume is becoming standard. However, it's not always a demonstrable skill that candidates can show in a portfolio.
At the same time, there is a user experience problem in the recruitment process: hiring managers are pressed for time and need to sift through static documents to find specific answers about a candidate's fit.
I started thinking about a solution that would solve the business inefficiency for recruiters, while serving as a functional proof of work regarding my ability to implement a Retrieval-Augmented Generation (RAG) tool in a corporate learning context.
Solution
Solution
The result is a functional, interactive AI Career Assistant that transforms my portfolio from a passive reading experience into an active engagement tool. Key capabilities include:
Instant Gap Analysis: Recruiters can paste a job description to receive an immediate, objective assessment of how my skills align with their specific role requirements.
Behavioral Interviewing: The agent can answer high-level questions (e.g., "How does Susan handle stakeholder conflict?") using real-world examples from my career history.
Proof of Competency: This tool moves AI from a theoretical bullet point on my resume to a tangible, live demonstration of my ability to adopt and deploy emerging learning technologies to solve real-world problems.
Process
Process
I approached this project as an instructional design challenge requiring a blend of prompting and technical implementation.
Knowledge Engineering and Curation: To move beyond a simple resume reader, I used the Dante AI platform to curate a specialized knowledge base that includes my resume, portfolio, and a selection of technical and behavioral interview questions that I have been asked in recent interviews, as well as my original answers in my own voice. This ensured the chatbot could answer higher level "how" and "why" questions with my authentic voice and specific examples. I also added my Workplace Insights results in order to respond to queries about my communication and interaction style, temperament, and work habits.
Technical Architecture (RAG): Leveraging Dante AI and GPT-4o mini, I curated a custom agent capable of retrieving specific data points to answer queries accurately.
Prompting: I designed a system persona (the professional talent recruiter/ hiring manager) to ensure the bot remained objective. I calibrated the personality and behavioral instructions to prioritize evidence-based answers over conversational filler, ensuring it adhered to professional standards.
Testing: I tested the bot with vague queries and complex job descriptions meant to refine its logic for the job fit analysis feature, ensuring it could accurately identify skill gaps and alignments without hallucinating qualifications that I don't have.
Page updated
Google Sites
Report abuse