Signal

In an era where AI has flooded recruiters with fabricated resumes. A 6-agent AI copilot with MINSKY cryptographic Git audit that verifies code provenance, auto-tracks applications from emails.

Comment

Problem Statement:

  1. Generative AI tools allow candidates to mass-produce fabricated resumes, prompting enterprise Applicant Tracking Systems (ATS) to deploy strict keyword filters that reject qualified candidates.
  2. Portfolios and resumes do not provide verifiable proof of authorship, leaving recruiters with no fast way to verify if a candidate authored their code or copied public repositories.
  3. Job applicants face significant administrative overhead manually maintaining spreadsheets, monitoring emails across multiple platforms, and scheduling interview follow-ups.

Who Faces the Problem:

  1. Software engineering students and job applicants facing automated ATS rejections and time-consuming manual application tracking.
  2. Technical recruiters and hiring managers who must filter through hundreds of AI-generated applications with unverified claims.
  3. University placement cells and career centers needing reliable, tamper-evident records of student technical competencies.

Solution:

  1. MINSKY Forensic Agent: Verifies code provenance by auditing GitHub repositories using cryptographic commit signatures (GPG/SSH), commit cadence, and Abstract Syntax Tree (AST) analysis.
  2. Email and Ingestion Agent: Ingests recruiter communications via Cloud Pub/Sub push streams, automatically extracting interview schedules and status changes.
  3. Career Optimization Agent: Conducts semantic gap analysis comparing verified candidate skills with job descriptions to identify missing qualifications.
  4. Tracking Agent: Synchronizes application statuses across a real-time Kanban board on Cloud Firestore.
  5. AI Drafting Agent: Uses Gemini 2.5 Flash to generate personalized cover letters and outreach referencing verified commit hashes and project metrics.
  6. Scheduled Nudge Agent: Schedules interview preparation reminders and recruiter follow-ups via Cloud Tasks.

What Makes It Unique:

  1. Provenance Verification: Validates authentic code authorship through cryptographic signatures and commit heuristics rather than relying on unverified resume claims.
  2. End-to-End Multi-Agent Automation: Operates an asynchronous event-driven workflow using Cloud Pub/Sub and Cloud Tasks to eliminate manual data entry.
  3. Evidence-Backed Outreach: Grounds candidate communications in verified commit metrics and tamper-evident scores rather than generic generative text.
  4. Production Cloud Architecture: Utilizes LangGraph, Gemini 2.5 Flash, Cloud Firestore, and Cloud Tasks for sub-second reactive updates and enterprise scalability.

Built With

Share this project:

Updates