#61
Monthly Rank
#163
Lifetime Rank
Share
11 September 2026
A

AlgoAudit India

Catch hiring bias before it reaches a candidate.

Sanchita Agrawal

Gallery

About the Project

AlgoAudit India is a pre-deployment audit tool that helps Indian employers check their hiring criteria and screening prompts for potential bias, privacy risks, and legal compliance concerns — before they're used to screen real candidates.

An employer enters a job description or the criteria they plan to use (either written by a person or fed into an AI screening tool). The system runs it through a hybrid analysis engine: a deterministic rule layer catches well-known problem patterns (age cutoffs, location filters, career-gap penalties, gendered or marital-status language, excessive personal data requests), and an LLM-based layer catches subtler or more novel phrasing. Together they flag potential risks, explain why each one is a concern, map it to relevant Indian legal/regulatory areas for the team to verify, and suggest a rewritten, job-relevant version of the criteria. A human reviewer then approves or edits the recommendation, and the whole audit — original criteria, findings, and final decision — is saved as a downloadable PDF report.

The problem it solves: hiring criteria and AI screening tools can unintentionally encode bias — an age cutoff, a "no career gap" rule, a location radius — that looks neutral on paper but disproportionately filters out qualified candidates from certain groups. Most of the time nobody catches this until it's already been used, or until it becomes a legal or reputational problem. AlgoAudit catches it at the design stage, with a transparent, explainable process instead of a black-box "reject" decision, and — importantly — it flags potential risk for human review rather than declaring anything definitively illegal, keeping a person in the loop for the final call.

Practice Areas

Key Features

•⁠ ⁠Multi-step audit submission — employer enters job title, optional industry context, and hiring criteria in a simple guided flow •⁠ ⁠Hybrid bias detection — deterministic rule-based pattern matching combined with LLM-based analysis for both consistency and depth •⁠ ⁠Transparent risk scoring — a point-based Low/Medium/High severity system computed from the actual flags raised, not an arbitrary AI-generated number •⁠ ⁠Explainable findings — every flag comes with a plain-language explanation and the exact excerpt from the criteria that triggered it •⁠ ⁠Indian legal/compliance reference mapping — flags are linked to relevant Indian laws and regulations for the team to verify before presenting as final •⁠ ⁠Before/after criteria comparison — original wording side-by-side with a suggested, job-relevant rewrite •⁠ ⁠Human review workflow — employer can approve or edit the recommendation; nothing ships without a human decision •⁠ ⁠Audit history dashboard — filterable by status and severity, with full past-audit detail on demand •⁠ ⁠Downloadable PDF report — a complete audit record including findings, legal references, and reviewer decision •⁠ ⁠Built-in compliance disclaimer and guardrails — the system is designed to flag potential risk, never declare a practice definitively illegal, with automated checks against overly definitive legal language

About the Creator

SA
Sanchita Agrawal