Best IT Training Institutes in Chennai with Placement | DeepNeuron

Job description

As a critical member of the team, your work will be cutting-edge technologies and will play a high-impact role in shaping the future of AI-driven enterprise applications. You will directly work with people whove worked at Amazon, Facebook, Google, and other technology companies in the world. With Level AI, you will get to have fun, learn new things, and grow along with us.

Roles and Responsibilities :

• Work on problems arising in speech-to-text pipelines, such as voice activity detection, transcription, automatic speech recognition (ASR) speaker diarization (SD),.
• Train, deploy and maintain scalable speech-to-text pipeline to power Level AI s ASR engine.
• Keep abreast with SOTA techniques in your area and exchange knowledge with colleagues.
• Work with other team members to develop architecture design of systems.
• Ability to independently conduct experiments with model architectures, training schemes, and approaches proposed in ASR literature.
• Work in an agile environment to deliver high-quality products.

Requirements :

• Bachelors in Computer Science or Electrical Engineering or related fields.
• Strong knowledge of Machine Learning fundamentals and Deep learning architectures like Transformer, Conformer etc
• Understanding of Classical Speech processing models like gaussian mixture models(GMM) and Hidden Markov models(HMM).
• Understanding of Connectionist temporal classification (CTC) and RNN-T objective functions.
• Hands-on building n-gram language model integrating with shallow fusion techniques.
• Hands-on experience with Python programming language and a Deep Learning framework like Pytorch/Tensorflow.
• Hands-on experience in deploying end-to-end speech recognition models using cloud applications like AWS, GCP.
• Awareness of state of the art research in speech recognition and Signal processing communities.
• Experience in Semi-supervised learning is a plus.
• Experience in building text-to-speech or punctuation restoration for ASR transcript models is a plus.