Job Title
Senior Applied Machine Learning Engineer – Speech & Language Systems (Confidential Project)
Location
Remote (Full-Time)
GCC Time Zone Preferred
About the Role
We are building a next-generation multilingual speech and language intelligence system designed for real-time, production-grade deployment.
This is a greenfield, R&D-intensive role for an engineer who has hands-on experience fine-tuning large speech and language models using modern parameter-efficient techniques.
We are specifically looking for someone who has built and deployed custom ASR, TTS, and LLM systems not someone who has only integrated APIs.
What You’ll Work On
- Fine-tuning state-of-the-art speech recognition models for dialect-heavy, real-world audio
- Training and attaching LoRA / PEFT adapters to transformer-based language models
- Improving intent understanding for noisy, informal, and code-switched language
- Fine-tuning multi-speaker TTS models for natural prosody and accent control
- Building low-latency inference pipelines for streaming audio systems
- Designing scalable data pipelines for large-scale speech dataset processing
- Optimizing quantized models for real-time deployment
Required Technical Experience
We are only considering candidates with hands-on experience in the following areas:
Speech Recognition (ASR)
- Fine-tuning models such as Whisper, wav2vec2, or similar architectures
- Adapter-based training (LoRA / PEFT)
- Training on domain-specific or dialect-heavy datasets
- Handling noisy, real-world audio
Large Language Models (LLMs)
- Supervised fine-tuning
- Adapter-based training
- Intent classification & slot extraction systems
- Retrieval-Augmented Generation (RAG)
- Model quantization (4-bit / 8-bit)
Text-to-Speech (TTS)
- Tacotron, VITS, XTTS, or similar architectures
- Multi-speaker training
- Accent and prosody adaptation
- Speaker embedding tuning
Infrastructure
- PyTorch
- HuggingFace Transformers
- Distributed training workflows
- Model serving optimization
- Real-time inference pipelines
Strong Plus
- Experience with dialectal or low-resource languages
- Experience building telephony-integrated speech systems
- Experience training on custom speech corpora
- Experience reducing ASR WER in non-standard speech environments
What This Role Is NOT
- Not a generic data science role
- Not an analytics or dashboard position
- Not prompt engineering only
- Not API-only AI development
This is a deep applied ML engineering role focused on model training, adaptation, and production deployment.
Ideal Candidate Profile
- 5+ years of ML engineering experience
- Has personally fine-tuned large-scale models
- Has deployed speech systems in production
- Comfortable with experimental R&D
- Strong mathematical and optimization background
- Thinks in terms of systems architecture, not demos
Compensation
Competitive salary + equity (depending on experience and impact potential).
Confidential project details shared after technical screening.
Job Type: Full-time