Senior Staff Engineer- ASR

Own and advance the complete ASR stack focused on accuracy, multilingual and code-switched Indian languages, streaming latency, and robustness in production audio.

Lead research and design for ASR modeling, select architectures, conduct experiments, and ship streaming and offline ASR systems meeting performance SLAs at enterprise scale.

Provide technical leadership by mentoring engineers, setting technical standards for ASR engineering, and representing the company's work at international conferences.

8-10 years experience in speech/ML research or engineering with significant contributions to production-grade ASR systems.

Deep expertise in multilingual ASR for Indian languages including code-switching, accents, dialects, and low-resource languages.

Strong hands-on experience with modern ASR architectures including Conformer, E-Branchformer, RNN-T, CTC, Whisper/encoder-decoder, and Speech Language Models.

Proficiency in PyTorch, NVIDIA NeMo, ESPnet, Hugging Face, Kaldi/k2, distributed training, large-scale evaluation, and production inference optimization tools like ONNX, TensorRT, and NVIDIA Triton.

Experienced individual contributor capable of setting technical direction and mentoring teams but with no people-management responsibilities.

Demonstrated ability to translate ASR research into enterprise-grade systems at scale, with deep understanding of real-world audio challenges like telephony, far-field, reverberation, and overlapping speakers.

Strong research presence evidenced by publications in major international conferences and familiarity with designing robust evaluation benchmarks.