EdgeTal is an AI-powered recruitment tool that runs entirely on your phone. Semantic search, candidate analysis, zero cloud dependency. Your data never leaves your device.
Natural language queries are converted into 512-dimensional semantic vectors using on-device MediaPipe embeddings. No keywords needed.
Cosine similarity matching against locally stored candidate profiles. When confidence is low, an agentic loop autonomously reformulates the query.
A quantised Gemma-2B model generates structured candidate assessments with explainable reasoning: Thought, Evidence, Conclusion.
All processing happens locally. No cloud servers, no API keys, no data transfer. GDPR and CCPA compliant by architecture, not by policy.
Tested on mid-range Android hardware. Semantic search completes in under 200ms, even on devices with 4GB RAM.
Exact keyword matches in yellow. Semantic concept matches in blue. See what traditional search misses at a glance.
EdgeTal's architecture is published in a peer-reviewed research paper presented at EICON 2026 (ESOFT International Conference). The system was empirically evaluated across two hardware tiers with benchmarks on retrieval accuracy, latency, scalability, and generative analysis quality.
EICON 2026 · ESOFT Uni, Sri Lanka · August 2026