Research & Documentation

An overview of Anthrolytic's mission, vision, and product direction for AI-powered candidate screening.

This academic work examines moral dimensions arising when ML models span the entire recruitment workflow. Rather than offering technical solutions, the paper concentrates on principles and potential harms relevant to practitioners when algorithms mediate employment access. The authors contend that ethical implementation demands persistent consideration of impartiality, openness, candidate respect, responsibility, data protection, human involvement, and regular evaluation.

We fine-tuned NVIDIA's PersonaPlex-7b-v1 model on 200 synthetic coffee shop customer conversations to address hallucinations and poor task adherence in customer-facing deployments. Using LoRA training with ChatterboxTTS audio and LibriSpeech voices, we introduced a semantic-weighted loss function and voice prompt injection mechanism to improve emotional accuracy and reduce role-inversion hallucinations. LLM-as-judge evaluation shows improvements over the base model across all three tested configurations.