- Location
- Boston, MA, United States · Remote
- Compensation
- Not stated in this snapshot
- Employment
- Full-time
- Experience level
- MID (2–5 years exp)
- Specialization
- Machine Learning Engineer, Wellfound
- Source
- Wellfound
FinOptima Solutions is developing real-time voice deepfake detection and AI interaction intelligence for banks and financial institutions. We have an existing model, research paper, datasets, GPU access and cloud infrastructure. We are now improving the model’s performance across different speakers, recording environments and telephone-quality audio.
**The role
We are looking for a hands-on Research Assistant who is currently pursuing a PhD in machine learning, speech processing, audio AI, signal processing, electrical engineering or a closely related field.
You will work with our CTO, machine learning team and academic advisor to improve the existing model and prepare it for controlled pilots. This is a practical research role involving code, datasets, experiments and model evaluation—not purely theoretical work.
What you will do
Research voice deepfake detection, audio provenance and anti-spoofing methods
Identify and prepare genuine and synthetic speech datasets
Process real-world and telephone-quality audio
Run training, fine-tuning and model-evaluation experiments
Investigate cross-domain and cross-speaker generalization issues
Analyze false positives and false negatives
Test preprocessing, calibration and audio-window aggregation methods
Maintain reproducible experiment logs and technical documentation
Support future academic papers and research submissions
Help translate research findings into practical product improvements
Requirements
Currently pursuing a PhD in a relevant technical discipline or already has a PhD
Hands-on experience with audio or speech machine learning
Strong Python skills and experience with PyTorch or TensorFlow
Understanding of signal processing, audio preprocessing and model evaluation
Ability to read academic papers and implement research methods
Organized, proactive and comfortable working independently
Preferred experience
Voice deepfake detection, anti-spoofing or speaker verification
ASVspoof, Common Voice or similar audio datasets
Access to university laboratory or GPU resources
Academic writing and publication experience
Interest in cybersecurity, responsible AI and financial services
This is an equity-only opportunity for a PhD candidate who wants to contribute to applied research with real commercial and societal impact. Publication authorship will be based on substantive research contributions.