Smit Marvaniya

CEO

Bengaluru, Karnataka, India12 yrs 5 mos experience

Most Likely To SwitchAI Enabled

Key Highlights

Led AI initiatives to prevent content abuse at LinkedIn.
Developed climate-aware demand forecasting models at IBM.
Pioneered intelligent tutoring systems using NLP and AI.

Stackforce AI infers this person is a leader in AI and machine learning across social media, climate tech, and edtech industries.

Contact

Skills

Core Skills

Artificial Intelligence (ai)LeadershipMachine LearningNatural Language Processing (nlp)Video Analytics

Other Skills

AlgorithmsCC#C++Chatbot DevelopmentComputer VisionData AnalysisData MiningData ScienceData StructuresDatabasesExecutive-level CommunicationGenerative AIJavaJavaScript

About

Smit lead the Content Abuse Prevention Team at LinkedIn. His team builds scalable models to detect and prevent content abuse on the LinkedIn platform. Smit possess extensive expertise in spearheading research and product development efforts to create groundbreaking solutions for diverse real-world applications. He was previously working as Global Research Lead at IBM Research, India where he was leading a global team of researchers and research engineers for developing a set of innovative capabilities in the area of Explainable AI, ML, CV, and NLP. Smit holds the Master of Science (M.S.) By Research degree in Computer Science and Engineering from the Indian Institute of Technology, Madras, India. Smit has published 15+ peer-reviewed papers in top-tier conferences/journals (CVPR, CIKM, NAACL, AIED, EDM, ECCV, ICIP, MVA, IJRS) and filed 55+ US patent applications, with 30 already granted. He is actively contributing to the professional community as a PC member in conferences and journals like AIED, IJCAI, ECML-PKDD, AAAI, MVA, and TJCV.

Experience

12 yrs 5 mos

Total Experience

3 yrs 1 mo

Average Tenure

4 yrs 3 mos

Current Experience

2 roles

Senior Manager, AI | Content Abuse | Trust

Apr 2025 – Present · 1 yr 1 mo

Engineering Manager, AI | Content Abuse | Trust

Jan 2022 – Mar 2025 · 3 yrs 2 mos

Strategic Leadership:
✔ Responsible for formulating comprehensive roadmaps, devising effective strategies, providing technical direction, and overseeing execution across diverse teams, including AI, product, legal, policy, and engineering.
✔ Leading a team of 15+ AI engineers, overseeing the execution of initiatives to prevent content abuse across various domains, including LinkedIn feed, Messaging, Ads, Jobs, etc., to ensure the safety of members from unprofessional content such as violence, hate speech, adult, fraud & scams, commercial/promotional spam, misinformation, etc.
Innovation in Content Moderation:
✔ Leading the Content Abuse Red Team at LinkedIn, I demonstrate leadership in developing and implementing advanced content moderation solutions optimized for quality and efficiency across policies, surfaces, content types, etc.
✔ Spearheaded the creation of a scalable content moderation platform utilizing novel GenAI technology for spam detection in text and multimedia signals.
Advanced Detection Models:
✔ Leading initiatives to construct and deploy state-of-the-art models for text, image, and video moderation, including the implementation of a deepfake model for LinkedIn profiles.
✔ Employed a diverse range of signals to significantly improve accuracy and efficiency in detecting and preventing content abuse on the LinkedIn platform, utilizing advanced multi-modal and multi-task frameworks.
✔ Leveraged zero-shot capabilities through the integration of LLM and LVM architectures, enabling effective multilingual and multimodal content classification for text, images, and videos.

Artificial Intelligence (AI)People ManagementTechnical LeadershipNegotiationMulti-disciplinary TeamsStrategic Thinking+9

Ibm research

2 roles

Global Research Lead | Advisory Research Scientist | Master Inventor

Promoted

Jun 2020 – Nov 2021 · 1 yr 5 mos

Climate-aware demand forecasting
✔ Responsible for leading cross-geographical teams spanning IBM Research UK, IBM Research Kenya, and IBM Research Bangalore. My role involves formulating a roadmap, strategy, technical directions, and orchestrating execution across multiple teams for a set of AI/ML use cases in the future of climate.
✔ Played a global research lead role for building & productizing climate-aware capabilities such as mid to long-term uncertainty-aware explainable demand forecasting models to enable resiliency in the manufacturing supply chain.

Artificial Intelligence (AI)Machine LearningNatural Language Processing (NLP)Data Analysis

Advisory Research Engineer

Jun 2016 – May 2020 · 3 yrs 11 mos

Watson Tutor (Multi-million-dollar IBM <> Pearson)
The goal of Watson tutor was to enable a dialog-based intelligent tutoring system using NLP and AI. The Waston Tutor was featured in the Discovery Channel documentary series called “This is AI”.
✔ Led research and development effort for student response analysis module that compares student responses with reference answers using NLP, ML, and DL-based techniques and provides feedback based on the classification of student response.
✔ Multiple research problems were solved by analyzing the data obtained from real student interactions with the AI chatbot.
Productised a technology to improve the performance of a model that combines the sentence embedding features with histograms of s partial similarity features for automatic short answer grading in dialog-based tutoring scenarios (AIED, 2018).
Productised a technology that helps in balancing human effort and performance of student response analyser in dialog-based tutors (AIED, 2018).
Productised a technology by formulating a scoring rubric that includes class-specific representative student answers to improve short answer grading (CIKM, 2018).
✔ Worked closely with the research team and product team and successfully productized key technologies.
✔ Granted 5 US patents and 7 Top tier papers (CIKM 2018, NAACL 2019, AIED 2018, EDM 2017, ICLS 2018)
Digital Farming Platform (Multi-million-dollar IBM <> YARA)
✔ Productised a scalable digital farming platform, securing a strategy multi-year partnership with YARA and IBM, targeting enhanced agriculture coverage across ~100M hectares.
✔ Led research and development efforts for field boundary delineation using the multi-stage approach that uses a combination of machine learning, computer vision and image processing techniques.
✔ Delivered the end-to-end analytics to IBM business unit (TWC) as a part of WDPA offerings.
✔ Filed 1 Patent and co-authored a journal paper (International Journal of Remote Sensing, 2020)

Natural Language Processing (NLP)Machine LearningData AnalysisArtificial Intelligence (AI)

Samsung electronics

2 roles

Lead Engineer

Promoted

Mar 2015 – Jun 2016 · 1 yr 3 mos · Bangalore

Video Analytics Solutions
✔ Designed and developed novel real-time video summarization, video stabilization, and video highlight solutions which were productized to Samsung flagship mobile phones and Samsung Magicinfo Solutions.
✔ Led and Mentored a team of research engineers.
✔ Collaborated closely with the product team and successfully productized the research assets.
Publication: Real-time video summarization on mobile, ICIP 2016
Patent: Method and apparatus for providing summary information of a video, 2017, US20170242554A1, IN1452CH2015

Video AnalyticsMachine LearningData Analysis

Senior Software Engineer

Jul 2014 – Feb 2015 · 7 mos · Bangalore

Part of Advanced Research Team (ART)

Oracle

Member of Technical Staff

Jul 2013 – Jul 2014 · 1 yr · Bangalore

Audit Vault and Database Firewall provides defense for databases and comprehensive monitor through consolidation of audit data from databases, directories, operating systems and file systems. A highly accurate SQL grammar-based engine monitors and blocks unauthorized SQL traffic before it reaches the database. Database activity data from the network is combined with detailed audit data for easy compliance reporting and alerting.
✔ Designed and Developed security features such as Oracle Database Vault into the AVDF product to address the most difficult security problems of protecting against insider threats, meeting regulatory compliance requirements, and enforcing separation of duties.
✔ Added the features to support ORACLE DATA GUARD and RAC(Real Application Clusters) into the AVDF product.
✔ Worked in C++/C components of the product along with migrating old C/C++ code into Java-based components for better usability.
✔ Added the feature using materialized view to reduce the time taken for the generation of the summary audit report.
✔ Added automated test cases to reduce manual testing.