emotion control, audio tokenizers, DeepSpeed, domain adaptation, voice cloning, large-scale pre-training, quantisation, speech-to-speech models, self-supervised pre-training, distributed training, speech-LLM pre-training, Megatron, Distillation, speech foundation models, codec design, Experimental Design, streaming ASR, FSDP, low-latency streaming TTS, GPU concurrency

Machine Learning, Shell Scripting, Deep Learning, Tensorflow, Pytorch, PERL, FastAPI, Python, AWS, PEFT, SpeechBrain, TDT, RNNT, vLLM, LORA, CTC, SpeechLM, nemo

Ml, Summarization, Nlp, Pytorch, Python, fine-tuning, versioning, structured clinical abstraction, Documentation, reproducibility, weak supervision, entity extraction, patient-trial matching, eligibility reasoning, Llm, tests, Active Learning, prompting, peer review, Normalization, RAG, hybrid ML

probability , Computer Vision, C++, Deep Learning, Cuda, API Design, Pytorch, Opencv, Image Processing, FastAPI, Python, Distributed Training, HuggingFace, Feature Extraction, TensorRT, Ray, Statistics, Nvidia Triton, Optimization, geometry, EdgeML, Model Deployment, Inference Pipelines, linear algebra

Pytorch, Python, Hugging Face, ESPnet, Model training, ASR fine-tuning, nemo, Evaluation frameworks, MLflow logging

Sql, Git, Pandas, Docker, Ffmpeg, Python, Airflow, Parquet, Prefect, Dagster, PyArrow, librosa, polars, JSONL, Sox, torchaudio