IEEE ICCNT 2026 · acceptedRecruitingMar 2025 - Jul 2025Project Lead
Mock'n-Hire
AI-powered resume screening & mock interviews
- Precision@10
- 86.4%Precision@10
- Top-5 accuracy
- 88.2%Top-5 accuracy
- emotion accuracy
- 82.6%emotion accuracy
- per 50 resumes
- 3.2minper 50 resumes
The problem
Recruiters skim hundreds of resumes with fixed templates and no control over what matters; candidates rehearse blindly, with zero feedback on how they come across under pressure.
What I built
Recruiter-defined weightage drives semantic ranking; a mock interview generates questions personalised to each resume; video responses are analysed for stress so candidates get instant, actionable feedback.
Highlights
- Modular FastAPI services: parsing, ranking, question generation, emotion analysis
- Mistral-8x7B + Sentence Transformers for recruiter-weighted semantic scoring
- MobileNetV2 stress analysis mapped to interview-specific feedback
- Evaluated on 250 resumes across 5 job roles
How it works
- 01Resume intakeBulk upload & parse
- 02Recruiter weightageWhat matters, set per role
- 03Semantic ranking
- 04Question generationPersonalised per resume
- 05Video response
- 06Stress & feedback