Full Feature Set — March 2026

Screen. Brief. Interview.
Fully automated.

HireSignal is the only platform that covers the entire pipeline — from GitHub scoring to an AI interviewer that runs the call for you.

Deterministic scoring — not LLM guessesAI interviews — zero recruiter timeNYC LL144 · EU AI Act · EEOC compliant
GitHub Signal Analysis
Feature 01

11-Signal Scoring Engine

A deterministic, explainable scoring model across eleven HR-relevant signals — each with a per-signal confidence band. Reproducible every time — no black boxes.

Profile Completeness — name, bio, location, links (max 10 pts)
Repository Quality — descriptions, topics, licensing (max 25 pts)
Community Impact — stars, forks & external OSS activity (max 10 pts)
Contribution Consistency — streak, diversity, cadence (max 20 pts)
Technical Breadth — language & topic diversity (max 15 pts)
Social Proof — followers, public presence (max 5 pts)
Commit Quality — message discipline & atomicity (max 20 pts)
Recency — how live is their activity right now? (max 10 pts)
PR Review Quality — external code review engagement (max 10 pts)
OSS Contributions — external project contributions (max 10 pts)
Documentation Quality — READMEs, descriptions, personal site (max 10 pts)
Feature 02

LLaMA 3.1 AI Analysis

A self-hosted LLaMA 3.1 8B model enriches every report with candidate-specific insights — not generic templates.

Plain-English executive summary for hiring managers
5 custom interview questions from the candidate's actual stack
Red flag detection (star-farming, account age mismatch, gaps)
Standout factor identification
AI-authored hire verdict with override transparency
Zero data sent to third-party LLM providers
Feature 03

Commit Quality Analysis

Goes beyond star counts — samples actual commit messages to assess engineering discipline and team habits.

Samples 50–100 commits across top repos
Scores message quality, length, and clarity
Detects conventional commit adoption (feat/fix/chore)
Flags patterns worth asking about: 'WIP', 'fix', 'asdf'
Separate commit quality score dimension
Contextual HR insight generated from commit patterns
Feature 04

Batch Screening

Screen an entire candidate pipeline at once. Rank, compare, and export — without touching a single repo manually.

Analyze up to 20 candidates simultaneously
Side-by-side score comparison table
Automatic ranking by overall score
One-click CSV export of full batch results
Per-candidate decision badges
Priority queue — see your top candidates first
AI Recruiter — 3-Stage Interview Pipeline
Pro · Enterprise

The full interview pipeline — AI-powered end to end

Three stages, two models, one coherent workflow. Claude Sonnet handles deep analysis where latency allows. LLaMA 8B handles real-time assistance where the candidate is live.

Stage 1

Interview Brief — Claude Sonnet

Before the recruiter joins a call, Claude Sonnet generates a deep pre-interview brief.

  • Reads every signal: GitHub, CV, LinkedIn, screening form
  • Ranks the 7 highest-signal questions by priority
  • Writes per-question rubrics: strong / adequate / weak answers
  • Identifies unverified claims worth probing
  • Suggests interview format (technical, portfolio, etc.) and number of rounds
  • All reasoning stored in the audit log (GDPR Art. 22)
1
Stage 2

Live Interview Co-Pilot — LLaMA 8B

The recruiter is in the meeting. The Co-Pilot assists in real time — silently, in the browser.

  • Listens as the recruiter types candidate answers
  • Suggests follow-ups in ≤200ms using LLaMA 8B
  • Scores each answer live against the brief rubric
  • Flags evasive or off-topic responses immediately
  • Notes key statements with one click
  • Summarises performance as the call progresses
2
Stage 3

Post-Interview Report — Claude Sonnet

When the interview ends, Claude Sonnet synthesises everything into a complete hiring report.

  • Hire / Hold / No-Hire recommendation with confidence
  • Question-by-question assessment against rubrics
  • Signal flags: strong answers, red flags, risk items
  • Calibrated hire recommendation with reasoning chain
  • Suggested next steps (technical test, reference, offer)
  • Fully exportable — PDF, ATS push, audit log entry
3

Claude Sonnet

Brief generation · Post-interview report

LLaMA 8B (self-hosted)

Real-time Co-Pilot · Autonomous turns

Zero third-party AI

No data sent to OpenAI or Anthropic

Autonomous AI Interviewer
Pro · Enterprise — Shipped March 2026

The AI conducts the interview.
You review the report.

Send a candidate a browser link. Mergen — HireSignal's AI interviewer — greets them, asks the brief questions, follows up intelligently, and closes the session when done. No recruiter is in the room. You receive the PostInterviewReport automatically when the session ends.

On-device STT

Browser Web Speech API — audio never transmitted

On-device TTS

Browser speech synthesis — Mergen speaks naturally

LLaMA 8B turns

≤200ms latency · self-hosted RunPod cluster

No audio stored

Only text transcript ever leaves the browser

Compliance built in: Mandatory AI disclosure on the consent screen (EU AI Act Art. 52 · NYC LL144). Candidate right-of-access transcript emailed automatically (GDPR Art. 15). Full audit trail for every session.

What the Autonomous Interviewer does

  • Recruiter sends a single link — AI conducts the full interview
  • Browser-native voice I/O (STT + TTS) — audio never leaves the device
  • Mergen (the AI) asks questions, follows up, and probes intelligently
  • Handles silence, empty responses, and evasive answers gracefully
  • Time-limit enforcement (server-side, per-session)
  • Candidate consent screen — EU AI Act Art. 52 & NYC LL144 compliant
  • PostInterviewReport auto-generated when session ends
  • Recruiter receives email notification — zero recruiter time required
  • GDPR right-of-access transcript emailed to candidate automatically
  • Full audit trail in Redis — 1-year retention for employment records

Where it sits in the pipeline

Stage 1 BriefCandidate opens linkAI InterviewStage 3 Report

The Autonomous Interviewer replaces Stage 2 when no recruiter is needed. It feeds directly into Stage 3.

And much more

Every edge case, handled.

One-Click Reports

Shareable PDF & link exports for hiring manager briefings. No engineering required.

Interview Guide

Auto-generated interview plan with rubrics, pipeline stage, and AI-enhanced questions.

ATS Export

Push candidate data directly to Greenhouse or Lever via webhook integration.

Pull Request Analysis

View the candidate's recent PRs across public repos — real collaboration signal.

Experience Inference

Automatically determines Junior / Mid-level / Senior / Staff level from public signals.

Role Suggestion

Recommends the best-fit engineering role based on language stack and project patterns.

Privacy-First Design

Self-hosted AI, public data only. History stored in your encrypted account — deletable anytime. No third-party LLM access.

Confidence Scoring

Every decision includes a data confidence %. Low public data triggers a 'needs review' flag.

Nuanced Hire Decisions

Strong Hire / Hire / Needs Review / Insufficient Data — never a blunt reject when data is sparse.

Under 60 Seconds

Full GitHub fetch, scoring, and AI narrative delivered before your coffee cools.

Gold Standard Benchmark

Compare any candidate against your own internal engineers to calibrate the bar.

AI Transparency Layer

When AI overrides the rule engine, a banner shows exactly what changed and why.

From screen to hire — all in one tool.

GitHub analysis · AI brief · live Co-Pilot · Autonomous Interviewer · post-interview report. Start free, no credit card.