Senior Machine Learning Engineer NLP, Speech & LLMs
Location: Coimbatore
We are looking for a Senior Machine Learning Engineer to design, build, and deploy core ML models powering an AI-driven SaaS experience platform. This role is ideal for someone who enjoys hands-on model development, working on NLP, Speech-to-Text (STT), Text-to-Speech (TTS), Speech Language Models, and LLM-based systems, and taking models from research to production.
You will work closely with researchers, product teams, and platform engineers to deliver scalable, production-grade ML systems.
Key Responsibilities
- Design, train, fine-tune, and evaluate ML models for NLP and speech use cases
- Build and optimize STT, TTS, and speech-language models
- Develop and experiment with transformer-based and deep learning architectures
- Prepare and manage large-scale text and speech datasets (cleaning, augmentation, labeling)
- Implement training pipelines, evaluation metrics, and benchmarking frameworks
- Optimize models for performance, latency, and cost in production
- Collaborate with engineering teams to deploy models as scalable services
- Monitor model performance and handle retraining and improvements
- Stay up to date with the latest research and apply it to real-world problems
- Mentor junior ML engineers and contribute to best practices
Required Skills & Qualifications
- 58+ years of experience in Machine Learning / Applied AI
- Strong hands-on experience in NLP and Speech Processing
- Proven experience building and training ML models (not just API usage)
- Solid experience with STT, TTS, and speech language models
- Strong understanding of deep learning fundamentals and transformer architectures
- Proficiency in Python and ML frameworks such as PyTorch (preferred) or TensorFlow
- Experience with libraries and toolkits such as Hugging Face, ESPnet, Fairseq, Kaldi, NeMo, Whisper (fine-tuning)
- Experience with LLMs, including fine-tuning, adapters (LoRA), and RAG pipelines
- Familiarity with GPU-based training, distributed training, and optimization techniques
- Strong grounding in statistics, linear algebra, and optimization
Preferred / Nice-to-Have
- Experience with self-supervised learning for speech or language (e.g., wav2vec, HuBERT)
- Knowledge of real-time or low-latency inference systems
- Experience with model compression, quantization, and distillation
- Exposure to MLOps tools (MLflow, Weights & Biases, Kubeflow, Airflow)
- Experience with cloud platforms (AWS, GCP, Azure)
- Open-source contributions or research publications
