Machine Learning Engineer Audio AI / Deepfake Detection with exeprience | EQUITY Only compensation before fundraising

About FinOptima FinOptima Solutions is building an AI trust layer for financial institutions. Our technology helps banks detect synthetic and manipulated voices during live customer interactions, before fraud results in financial loss. We have an existing voice deepfake detection model, research foundation, cloud infrastructure and an early demo. We are now strengthening the model’s performance on real-world and telephone-quality audio and preparing the product for controlled pilots with financial institutions. The role We are looking for an experienced, hands-on Machine Learning Engineer specializing in audio, speech or signal processing. You will take ownership of evaluating and improving our existing model and help move it from research into a reliable pilot-ready product. What you will do Review the existing model, training pipeline, datasets and evaluation methodology Diagnose cross-domain and generalization issues Improve performance on real-world and telephone-quality audio Reduce false positives on genuine human voices Evaluate retraining, fine-tuning, preprocessing and threshold calibration strategies Improve audio-window aggregation and full-call scoring Build a reliable validation framework using metrics such as EER, FPR, FNR, precision and recall Work with the CTO and research team to integrate the model into a real-time demo Document technical decisions, experiments and results What we are looking for Strong experience in machine learning for audio, speech or acoustic signal processing Hands-on experience with Python and PyTorch or TensorFlow Experience training, fine-tuning and evaluating deep-learning models Understanding of domain shift, model generalization and calibration Familiarity with speech embeddings, transformers, CNNs or related audio architectures Ability to independently review an existing codebase and research work Strong experimental discipline and clear technical communication Experience with voice anti-spoofing, deepfake detection or speaker verification Knowledge of ASVspoof, Common Voice or similar speech datasets Experience with telephony audio, codecs, Twilio or real-time inference Experience deploying ML models on GCP Research publications in speech, audio AI or signal processing This is an equity-only opportunity for someone who wants to join at an early stage, take meaningful technical ownership and help build a category-defining AI security product for financial institutions.