Sonam Damani

AI Researcher

Bellevue, Washington, United States12 yrs 5 mos experience

Most Likely To SwitchHighly Stable

Key Highlights

Led innovative AI projects at Microsoft.
Expert in Deep Learning and Natural Language Processing.
Awarded for excellence in innovation and execution.

Stackforce AI infers this person is a Deep Learning and NLP expert in the AI industry.

Contact

sonamdamani@gmail.com LinkedIn

Skills

Core Skills

Machine LearningQuery UnderstandingBing Web Search RankingDeep LearningNatural Language Processing (nlp)Software DevelopmentData Analysis

Other Skills

.NET FrameworkAlgorithmsAndroidAzure servicesBERTBiLSTMCC#C++ClassificationConcept driftConversational bot developmentCortana designCross-team CollaborationDashboard creation

About

Software and AI enthusiast with a fondness for innovative and challenging projects. Currently working on advancing Deep Learning applications for improving relevance of Microsoft Ads. In past, worked on enabling conversational capabilities for Microsoft Ruuh (https://www.ruuh.ai/) and Microsoft QnAMaker (https://www.qnamaker.ai/), building personality engine for Microsoft Cortana, topic prediction and relevance improvements for Bing search, query understanding for Bing search, novel art generation using AI, comment generation for images, and built dashboards, tools and services for monitoring and analytics. Experienced in Machine Learning, Deep learning, NLP, Data Mining, Data Wrangling, Statistics, Analytics, Azure, Web Application Development.

Experience

Microsoft

7 roles

Principal Applied Scientist Manager

Promoted

Sep 2022 – Present · 3 yrs 6 mos

Senior Applied Scientist

Jun 2020 – Aug 2022 · 2 yrs 2 mos

Worked on improving relevance for Microsoft text ads using ML techniques and SOTA models, while leading efforts for query understanding, measurement and monitoring, and syndication relevance.

Machine LearningSOTA modelsQuery understandingMeasurement and monitoringSyndication relevance

Senior Applied Scientist

Promoted

Sep 2019 – Jun 2020 · 9 mos

Led efforts for Bing web search ranking improvements by building topic-prediction models trained to learn concept drift in multi-task setting. Enabled existing rankers to leverage richer information from documents with input length and latency constraints.

Bing web search rankingTopic-prediction modelsConcept driftMulti-task settingMachine Learning

Applied Scientist II

Promoted

Aug 2016 – Aug 2019 · 3 yrs

Built Question Answer ranker using cutting edge deep learning models like BERT and MT-DNN. Worked on model compression techniques like knowledge distillation for Microsoft QnAMaker. Was awarded "Excellence Award for Innovation" for building innovative solutions for QnAMaker and FAQ Answering. Worked on adding a BiLSTM Sequence to Sequence response generation model in Ruuh. Enabled image understanding in Ruuh, and integrated an innovative technique for responding to images with natural comments (https://blogs.msdn.microsoft.com/msind/2018/07/12/ruuh-ai-image-recognition/). Created a transfer-learning based model for sensitive content detection in images. Trained GAN models coupled with optimization techniques to use AI for designing jewelry and painting images. Built the official Ruuh website (https://www.ruuh.ai/) with full fledged chat interface.

Deep LearningBERTMT-DNNKnowledge distillationBiLSTMGAN models+2

Software Engineer II

Feb 2016 – Jul 2016 · 5 mos

Among the founding members for building Ruuh, an open-ended conversational bot (https://m.me/Ruuh). Built capabilities for handling of offensive messages from users. Enabled several skills and scenarios for Ruuh including utilitarian features and interactive games. Built a dashboard for deriving KPIs and other conversation-associated metrics from user chat logs. Built several tools and Azure services in the process. Was awarded "Microsoft Spot Award" for full-fledged execution of various tasks.

Conversational bot developmentDashboard creationAzure servicesSoftware DevelopmentNatural Language Processing (NLP)

Software Engineer

Jun 2013 – Jan 2016 · 2 yrs 7 mos

Worked on single-class and multi-class classification for identifying intents like Image, Video, TV-show from Bing queries. Designed and implemented generic experience infrastructure for Cortana. Also worked on building next generation design for Cortana Personality to improve quality and efficiency. Was awarded "Excellence Award for Execution" for consistent performance and impact.

ClassificationCortana designIntent identificationNatural Language Processing (NLP)Software Development

Software Engineering Intern

May 2012 – Jun 2012 · 1 mo · Hyderabad Area, India

Worked on improving Bing Search Relevance for Tech Segment through Query Reformulation. Involved data mining and pattern analysis to build algorithm for finding queries of similar intent.

Query reformulationData miningPattern analysisData Analysis

Bsnl telecom factory

Data Analytics Intern

May 2011 – Jun 2011 · 1 mo · Bhopal Area, India

Developed descriptive statistics from customer queries and complaints and applied the insights to build a tool for minimizing handling cost thereby maximizing business value.

Descriptive statisticsTool developmentData Analysis