Prabhat Pandey

AI Researcher

Bengaluru, Karnataka, India11 yrs 7 mos experience

Most Likely To SwitchAI Enabled

Key Highlights

Expert in Automatic Speech Recognition and Natural Language Processing.
Led significant projects at Amazon and Microsoft.
Released highly downloaded datasets in speech/audio.

Stackforce AI infers this person is a specialist in AI-driven speech recognition technologies.

Contact

prabhatp251@gmail.com LinkedIn

Skills

Core Skills

Machine LearningSpeech RecognitionAutomatic Speech RecognitionRnn-t ArchitecturePredictive Modeling

Other Skills

Artificial IntelligenceBig DataC#Cognitive ServicesData AnalysisDataset ManagementHMM-DNN/LSTM modelsHybrid ASR modelingLattice-attentionNatural Language ProcessingPythonQuery-Entity Matching

Experience

11 yrs 7 mos

Total Experience

5 yrs 9 mos

Average Tenure

8 yrs 7 mos

Current Experience

Amazon

2 roles

Senior Applied Scientist

Promoted

Jan 2022 – Present · 4 yrs 5 mos · Bengaluru, Karnataka, India

Part of the core team that initiated the Nova Sonic project, a real-time conversational speech model.
Speech Language Models.
Self-learning in spoken dialog systems.
Released SIFT-50M dataset on Hugging Face – among the most downloaded in speech/audio category.

Machine LearningSpeech RecognitionNatural Language ProcessingArtificial Intelligence

Applied Scientist II

Sep 2017 – Nov 2021 · 4 yrs 2 mos · Aachen, North Rhine-Westphalia, Germany

Worked in the automatic speech recognition (ASR) team within Alexa AI organization.
Our team delivered ASR models for Alexa launches in multiple countries — Germany, UK, France, Italy, Spain and Brazil (Portuguese).
Lattice-attention based second-pass rescoring for on-device ASR systems.
End-to-end streaming ASR modeling with RNN-T architecture.
Open-vocabulary hybrid ASR modeling for agglutinative and compound-rich languages.

Automatic Speech RecognitionRNN-T architectureLattice-attentionHybrid ASR modeling

Microsoft

3 roles

Applied Scientist

Promoted

Feb 2016 – Aug 2017 · 1 yr 6 mos

Worked in the Speech team within Microsoft's AI & Research organization.
Our team developed ASR models for Indian English and Hindi, integrated into products like Cortana, Microsoft Translator, and Cognitive Services.
Built HMM-DNN/LSTM acoustic models and n-best rescoring models for hybrid ASR systems.

Automatic Speech RecognitionHMM-DNN/LSTM modelsCognitive Services

Software Engineer

Jul 2014 – Jan 2016 · 1 yr 6 mos

Delivered pre-game prediction model for 2015 Cricket World Cup which featured on live television broadcasts (Star Sports) and Bing Search.
Prototyped in-game predictions which first featured on Bing during the 2016 T20 World Cup, well before it became mainstream in cricket broadcasting.

Predictive ModelingData Analysis