Research Scientist, Voice Infrastructure
Lead research on speech recognition and synthesis for low-resource African languages, the science behind Voxtra and the voice layer of Luso8 Cloud. Own the path from published research to models running in real-time telephony.
Location
Lilongwe, Malawi (relocation supported)
Type
Full-time · Senior
Compensation
Competitive + equity
Responsibilities
- 01Design and train ASR and TTS models for low-resource, multilingual, and code-switched African-language speech.
- 02Push the accuracy/latency frontier for streaming inference under real telephony constraints (line noise, accents, sub-second turnaround).
- 03Own the research agenda for the Voice AI thread end to end: problem framing, experiments, evaluation, publication.
- 04Partner with the Voxtra and Luso8 engineering teams to move models from research checkpoints into production inference.
- 05Publish findings (top-tier venues or technical notes) and represent the lab’s voice research externally.
- 06Mentor engineers and junior researchers working on adjacent speech and NLP problems.
Requirements
- 01PhD in Computer Science, Electrical Engineering, Computational Linguistics, or a closely related field, with a dissertation or research focus on speech/audio ML (exceptional candidates with equivalent industry research experience will be considered).
- 02Strong publication record in speech processing or related ML venues (e.g. INTERSPEECH, ICASSP, ACL, NeurIPS).
- 03Deep hands-on experience with modern ASR/TTS architectures (e.g. Whisper-style encoders, Conformer, wav2vec 2.0, VITS, HiFi-GAN, or comparable sequence models).
- 04Fluency in PyTorch (or equivalent) and large-scale model training, including data pipelines for low-resource languages.
- 05Experience with streaming/real-time inference under production latency budgets.
- 06Working knowledge of telephony or voice-infrastructure concepts (SIP, RTP, WebRTC) is a plus, not a requirement.
- 07Prior experience with African languages, low-resource NLP, or code-switching is a strong plus.
- 08Comfortable operating as the senior voice-research authority in a small, fast-moving lab.
Benefits
- 01Define and lead the lab’s voice-research agenda from first principles.
- 02Direct line from research to production: your models ship in Voxtra and Luso8 Cloud, not just a paper.
- 03Potential equity stake of up to 2-5%, based on role, seniority, and impact (standard 4-year vesting, 1-year cliff).
- 04Publication support and conference travel for peer-reviewed work.
- 05Competitive salary, relocation support to Lilongwe.
- 06Collaborative, low-meeting environment built for deep research work.
Application process
- 01Email careers@rexplore.ai with subject "Research Scientist, Voice Infrastructure - [Your Name]".
- 02Attach: CV, publication list, and links to any released models, code, or demos.
- 03A short research statement (half a page) on a voice/speech problem you’d want to work on here.
- 04Round 1: research conversation with the Voice AI thread lead.
- 05Round 2: technical deep-dive on a past project or a take-home research problem.
- 06Final discussion about scope, start date, and offer.
Ready to apply?
We respond within two weeks. If you do not see a role that fits, send a fellowship inquiry, we hire ahead of named openings.