The world's first truly autonomous recruitment platform that eliminates human bias through deep reasoning and biometric engagement analysis.
PRISM is built on the premise that recruitment is fundamentally broken: biased, unscalable, and expensive. Our protocol replaces superficial keyword matching with Deep Reasoning. We understand not just what a candidate knows, but how they think.
- π‘οΈ Zero Bias: Automated PII obfuscation evaluating candidates purely as anonymous skill vectors.
- π§ Cognitive Intelligence: Chain-of-Thought reasoning for nuanced technical assessment.
- ποΈ Sentience Tracking: High-frequency biometric analysis to measure genuine candidate engagement.
- 𧬠Semantic Resonance: 1536d vector alignment between talent and opportunity.
PRISM leverages a state-of-the-art, multi-modal AI stack for total recruitment autonomy:
- Engine: Ollama
- Models:
DeepSeek-R1,Phi-3.5,Llama-3.2 (3B) - Role: Conducts L8 Chain-of-Thought (CoT) evaluation, generates adaptive interviewer prompts, and performs deep logic verification on coding responses.
- Model:
Sentence-Transformers (SBERT / BGE-M3) - Vector DB:
LanceDB(Serverless, Disk-based) - Role: Transforms resumes and job descriptions into high-dimensional embeddings. It identifies "non-obvious" fit patterns (e.g., recognizing that a 'Distributed Systems' expert is a fit for a 'Cloud Architect' role).
- Framework:
MediaPipe+OpenCV - Landmarks: 468-point 3D Facial Mesh tracking at 60Hz.
- Role: Monitors cognitive load, engagement variance, and authenticity during interviews. Detects proctoring violations (e.g., unauthorized devices) via
YOLOv8n.
- ASR (Speech-to-Text):
Faster-Whisper (Large-v3)&Moonshine-Tiny - TTS (Text-to-Speech):
Kokoro-v0.19(High-fidelity neural voice) - Role: Provides a natural, human-like verbal interaction experience with sub-200ms transcription latency.
Managed via Docker Compose, the system is orchestrated into specialized microservices:
| Service | Technology | Description |
|---|---|---|
| Gateway | FastAPI | Asynchronous high-performance API entry point. |
| Worker Grid | Celery + Redis | Handles heavy AI tasks (Transcription, Model Inference). |
| UI Hologram | React + Framer Motion | Premium 'Void Black' interface with holographic components. |
| Cognitive DB | PostgreSQL | Relational storage for candidate metadata and scoring. |
| Vector Index | LanceDB | Low-latency semantic search and retrieval. |
| Inference Node | Ollama | Local hosting of GGUF-based Reasoning Models. |
Docker ensures all C++ and AI dependencies (MediaPipe, SBERT, ONNX) are perfectly linked.
# Clone the repository
git clone https://github.com/AmanTShekar/PRISM-Augmented-Recruitment-System.git
cd PRISM
# Build the ecosystem
docker-compose up --buildAs AI models are large (>500MB), they must be pulled after the services start:
- Download Local Models (ONNX/Voices):
docker-compose exec backend python app/scripts/download_voice_models.py - Pull Reasoning Models (Ollama):
docker-compose exec backend python app/scripts/pull_models.py
- Autonomous Screener: JD entry β Auto-generated criteria β 24/7 AI Interviewing.
- Sentinel Identity: 180-degree face verification and liveness detection.
- Reasoning Reports: Generates full CoT logic traces for every hiring recommendation.
- Holographic Resumes: Interactive, AI-enhanced candidate profile visualizations.
- Proctoring Suite: Multi-person detection and screen-sharing integrity verification.
- Q1: Foundation - Core bias-free screening & Sentinel v1.
- Q2: Intelligence - DeepSeek-R1 full integration & Multi-language support.
- Q3: Scale - API v1 and Enterprise ATS connectors (Greenhouse/Lever).
- Q4: Platform - White-labeling for recruitment agencies and predictive analytics.
AI Recruitment System, Autonomous Hiring Platform, Deep Reasoning Interviewer, Sentinel AI Proctoring, Sentence-Transformers Matching, Machine Learning Recruitment, Zero Bias Hiring Protocol.
This project is licensed under the MIT License. See the LICENSE file for details. For commercial inquiries, contact [aman.t.shekar03@gmail.com].
Built with β€οΈ by the PRISM Team
"Defining the next era of autonomous recruitment."