Building the future of home entertainment(Successful, but confidential, exit)

(Via acquisition of Semantic Machines)My primary focus was Conversational AI. Specifically, I was working on deep learning models for spoken conversational behaviors such as turn-taking.

Acquired by Microsoft

Worked on several projects including: IARPA Babel (ASR & keyword spotting for low/under-resourced languages), ASR on mobile devices, deep neural nets for acoustic modeling, delay-and-sum beamforming using GCC-PHAT, integrating the Framenet lexical semantic database with NLTK.

Participated in the JHU HLT/COE summer (SCALE) workshops. 2009: worked on zero-resource ASR, 2013: worked on domain-mismatch issues within an ivector/PLDA speaker id system.

I led a team of research engineers working on both spoken (ASR+NLP) and text-only (NLP) dialog systems for the intelligent personal assistant space (web and mobile). We focused on a wide variety of topics including: tight integration of our wfst-based ASR decoder with our backend NLU system, incorporating contextual information (situational and conversational) into our ASR/NLP systems, statistical modeling for question-answering, human-to-human conversational analysis, visualization and understanding of large quantities of micro-texts, tools for enhancing the performance of language analysts, and machine translation.

I taught Intro to Computer Science, and selected lectures on Natural Language Processing and Speech Recognition.

Mainly focused on speaker diarization within the meeting domain and on ASR. Briefly worked on the Framenet lexical semantic database.

Research in the area of Human Language Technology. Acted as the interface between BBN's speech and NLP groups and the DoD.

Performed research into Human Language Technology.

Created an FDA-approved automatic speech recognition (ASR) system for the AESOP surgical robotic arm using the hidden Markov Model toolkit (HTK).

Research into Human Language Technology.