Work › AI Recruitment Platform
AI Recruitment Platform
A recruitment platform for students, recruiters and colleges, with resume scoring, AI-led mock and company interviews, and candidate matching.
This AI Recruitment Platform case study shows how Kynyx Solutions delivered LLM & generative AI solutions, agentic AI solutions and MVP development for startups for a HR tech / ed tech business using Next.js, FastAPI, MongoDB and faster-whisper.
Industry: HR Tech / Ed Tech · Year: 2026 · Visit live site
Result: AI interviews, resume scoring and candidate matching, with video and resumes kept away from the language model
Overview
The platform connects three groups: students preparing for jobs, recruiters hiring them, and colleges tracking placements. Students upload resumes and practise with AI-led interviews, recruiters run recorded company interviews and see matched candidates, and colleges get placement analytics.
The challenge
Screening large student pools by hand is slow, and generic AI tools raise fair questions about privacy and bias. The client needed interviews that adapt to each job, scoring that recruiters can understand, and a design that keeps personal data out of third-party models.
The solution
Interviews are generated one question at a time from the job description, and each question is checked for relevance and for protected topics, with a fixed fallback if the language model is unavailable. Speech is transcribed locally with faster-whisper, resumes are parsed and given an ATS score, and candidates are matched on a 70/30 mix of skills and ATS score. Video, resumes and face data never go to the language model, and face-continuity checks confirm the same person stays on camera.
Process
- Matching model — Resume parsing and ATS scoring, with matching weighted 70% on skills and 30% on ATS score.
- Adaptive interviews — Questions generated from the job description, checked for relevance and protected topics, with a deterministic fallback.
- Privacy and identity — Local transcription, no video or resumes sent to the language model, and face-continuity checks during interviews.
- Large uploads — Interview videos upload straight to the FastAPI backend to avoid serverless size limits.
Results
- 70/30 skills to ATS score in candidate matching
- 0 videos or resumes sent to the language model
- 3 user groups: students, recruiters, colleges
Services & technology
- LLM & Generative AI Solutions
- Agentic AI Solutions
- MVP Development for Startups
- Next.js
- FastAPI
- MongoDB
- faster-whisper
- OpenCV
- Groq
- PyMuPDF