AI Recruiting Agent
š¤
5
Bias-aware RAG AI recruiting agent with hallucination checks
Arjun Singh PRO
19arjun89
AI & ML interests
None yet
Recent Activity
updated
a Space
about 9 hours ago
19arjun89/AI_Recruiting_Agent
updated
a dataset
about 10 hours ago
19arjun89/ai_recruiting_agent_usage
updated
a Space
2 days ago
19arjun89/AI_Recruiting_Agent_Usage
Organizations
updated
a
Space
about 9 hours ago
updated
a
dataset
about 10 hours ago
updated
a
Space
2 days ago
published
a
Space
2 days ago
Great Work & Related Space
#1 opened 5 days ago
by
19arjun89
Great Web Page + Related Space
#1 opened 5 days ago
by
19arjun89
Great Work + Related Space
#1 opened 5 days ago
by
19arjun89
liked
2
Spaces
5 days ago
Great Work & Related Space
#1 opened 5 days ago
by
19arjun89
posted
an
update
5 days ago
Post
3419
Update: Making My AI Recruiting Assistant More Deterministic, Auditable, and Bias-Aware
Hi everyone ā I wanted to share a progress update on my AI recruiting assistant and some recent changes focused on reliability and transparency.
The goal of this project is to build a decision-support tool for recruiters that doesnāt just āsound confident,ā but can actually explain why it produces a given recommendation.
Link: 19arjun89/AI_Recruiting_Agent
Over the last few iterations, Iāve focused on three areas:
1) Deterministic Verification of Job Requirements (Skills)
Previously, required skills were extracted by an LLM from the job description. While this worked well, it still relied heavily on model behavior.
Iāve now added a verification layer that:
Requires every ārequiredā skill to be backed by a verbatim quote from the job description
This means hallucinated skills are explicitly detected and removed before scoring.
The system now shows:
What the model extracted
What was verified
What was dropped
Why it was dropped
This makes the requirements pipeline auditable instead of opaque.
2) Evidence-Based and Weighted Culture Verification
Culture matching is usually where AI systems become vague or subjective.
Iāve reworked this part so that:
Culture attributes are framed as observable, job-performance-related behaviors (e.g., audit readiness, operational reliability, security rigor)
Each matched attribute must include verbatim resume evidence
Matches are classified as:
Direct evidence (full weight)
Inferred evidence (partial weight)
Scoring is now weighted:
Direct = 1.0
Inferred = 0.5
This prevents āvibe-basedā culture scoring and makes the math transparent.
The output now shows:
The weights used
Which attributes were supported directly vs inferred
Which attributes were missing
3) Improved Bias Audit Prompt
Iāve also upgraded the bias audit prompt to be more structured and actionable.
Hi everyone ā I wanted to share a progress update on my AI recruiting assistant and some recent changes focused on reliability and transparency.
The goal of this project is to build a decision-support tool for recruiters that doesnāt just āsound confident,ā but can actually explain why it produces a given recommendation.
Link: 19arjun89/AI_Recruiting_Agent
Over the last few iterations, Iāve focused on three areas:
1) Deterministic Verification of Job Requirements (Skills)
Previously, required skills were extracted by an LLM from the job description. While this worked well, it still relied heavily on model behavior.
Iāve now added a verification layer that:
Requires every ārequiredā skill to be backed by a verbatim quote from the job description
This means hallucinated skills are explicitly detected and removed before scoring.
The system now shows:
What the model extracted
What was verified
What was dropped
Why it was dropped
This makes the requirements pipeline auditable instead of opaque.
2) Evidence-Based and Weighted Culture Verification
Culture matching is usually where AI systems become vague or subjective.
Iāve reworked this part so that:
Culture attributes are framed as observable, job-performance-related behaviors (e.g., audit readiness, operational reliability, security rigor)
Each matched attribute must include verbatim resume evidence
Matches are classified as:
Direct evidence (full weight)
Inferred evidence (partial weight)
Scoring is now weighted:
Direct = 1.0
Inferred = 0.5
This prevents āvibe-basedā culture scoring and makes the math transparent.
The output now shows:
The weights used
Which attributes were supported directly vs inferred
Which attributes were missing
3) Improved Bias Audit Prompt
Iāve also upgraded the bias audit prompt to be more structured and actionable.
posted
an
update
10 days ago
Post
138
š Update: One-Click Demo Mode Added to AI Recruiting Agent
Iāve added a one-click demo mode to the Candidate Assessment tab of my Hugging Face Space: AI Recruiting Agent.
š 19arjun89/AI_Recruiting_Agent
š Whatās new
You can now experience the full recruiting workflow end-to-end with a single click, without uploading any files.
The One-Click Demo automatically:
Clears any existing vector data
Loads sample company culture documentation
Loads a sample candidate resume
Loads a sample job description
Runs the full candidate evaluation pipeline
All using sample documents included directly in the Space repo.
š§Ŗ What the demo showcases
Skills match analysis
Culture fit evaluation
Final hiring recommendation
Hallucination verification & self-correction
Bias audit (pedigree bias, subjective language, non-traditional paths)
This makes it much easier for first-time users, reviewers, and evaluators to understand what the system does and how responsible AI safeguards are appliedāwithout setup friction.
š”ļø Responsible by design
The demo uses the same safeguards as real usage:
Resume anonymization (no PII in embeddings or prompts)
Fact verification against source documents
Bias audit chain surfaced to the user
Human-in-the-loop framing (decision support, not automation)
If youāre exploring AI-assisted hiring workflows or interested in transparent, bias-aware evaluation pipelines, Iād love feedback.
Iāve added a one-click demo mode to the Candidate Assessment tab of my Hugging Face Space: AI Recruiting Agent.
š 19arjun89/AI_Recruiting_Agent
š Whatās new
You can now experience the full recruiting workflow end-to-end with a single click, without uploading any files.
The One-Click Demo automatically:
Clears any existing vector data
Loads sample company culture documentation
Loads a sample candidate resume
Loads a sample job description
Runs the full candidate evaluation pipeline
All using sample documents included directly in the Space repo.
š§Ŗ What the demo showcases
Skills match analysis
Culture fit evaluation
Final hiring recommendation
Hallucination verification & self-correction
Bias audit (pedigree bias, subjective language, non-traditional paths)
This makes it much easier for first-time users, reviewers, and evaluators to understand what the system does and how responsible AI safeguards are appliedāwithout setup friction.
š”ļø Responsible by design
The demo uses the same safeguards as real usage:
Resume anonymization (no PII in embeddings or prompts)
Fact verification against source documents
Bias audit chain surfaced to the user
Human-in-the-loop framing (decision support, not automation)
If youāre exploring AI-assisted hiring workflows or interested in transparent, bias-aware evaluation pipelines, Iād love feedback.
Roadmap + Weekly Updates for AI Recruiting Agent (Feedback Welcome)
1
#1 opened 16 days ago
by
19arjun89
replied to
their
post
14 days ago
Some technical details for those interested:
- Resume and culture docs are chunked into Chroma vector stores
- Candidate scoring runs in two independent chains (skills + culture)
- Each analysis is fact-checked against source inputs
- Unverified claims can trigger a self-correction routine
- A separate bias audit prompt triangulates across skills, culture, and the final recommendation
Everything is designed as decision support ā not an automated hiring gate.
replied to
their
post
14 days ago
š Feedback Welcome
Iād love input on:
ā¢ Fairness pitfalls or bias signals I may have missed
ā¢ Edge cases in resume anonymization
ā¢ Verification or audit patterns youāve seen work well
ā¢ Features that would make this more useful in real hiring workflows
Happy to answer technical questions or iterate based on feedback.
published
a
dataset
15 days ago
Related Space and Feedback Sharing
#1 opened 16 days ago
by
19arjun89
General Feedback and Related Space
#1 opened 16 days ago
by
19arjun89
Great Work and Related Space
#3 opened 16 days ago
by
19arjun89
liked
a
Space
20 days ago