Kanchan Sarkar — Engineering Manager

Kanchan Sarkar is an Engineering Manager at TikTok, where he leads a global AI research and engineering team across Singapore, Beijing, Shanghai, and San Jose. His team focuses on building large-scale AI systems and high-performance models that power TikTok’s Content Intelligence platform. His team supports several critical downstream applications on the platform, including content diversity, content similarity detection, unoriginal content identification, deduplication, content safety, copyright protection, and the Creator Rewards Program. From a technical perspective, his team develops and deploys advanced Large Language Models (LLMs), Vision-Language Models (VLMs), and multimodal (“Omni”) models capable of understanding multiple modalities available on the platform, including text, images, video, and audio. The team pre-trains internal foundation models and adapts them for large-scale production deployment across a wide range of downstream applications. Kanchan brings deep expertise in: Large Language Models (LLMs) Multimodal Learning and Vision-Language Models (VLMs) Omni-modal AI systems (text, image, video, audio) Multimodal representation learning (e.g., VLM2Vec) Web-scale text, image, and video retrieval systems Visual similarity, content deduplication, and semantic matching Recommendation systems and content integrity With a strong background spanning AI research and large-scale production systems, Kanchan leads initiatives that bridge foundational AI research with real-world applications at global scale.

Stackforce AI infers this person is a leader in AI-driven solutions for social media and e-commerce industries.

Location: San Jose, California, United States

Experience: 9 yrs 8 mos

Skills

Large Language Models (llm)
Multimodal Learning

Career Highlights

Led global AI teams across multiple locations.
Developed advanced AI models for content intelligence.
Published research in top-tier AI conferences.

Work Experience

TikTok

Engineering Manager - TikTok AI (Content Intelligence) (2 mos)

Manager - TikTok AI (Content Intelligence) (2 yrs 2 mos)

Tech Lead (AI-Lab) (1 yr 10 mos)

Senior Research Engineer (AI-Lab) (1 yr 9 mos)

Shopee

Senior Data Scientist (1 yr 5 mos)

DataWeave Pvt. Ltd.

Data Scientist (Data Semantics) (1 yr 11 mos)

Riverbed Technology

Member Of Technical Staff (7 mos)

Education

Master of Technology (M.Tech.) at Indian Institute of Technology, Bombay

Bachelor of Engineering (BE) at Indian Institute of Engineering Science and Technology (IIEST), Shibpur

Kanchan Sarkar

Engineering Manager

San Jose, California, United States9 yrs 8 mos experience

Most Likely To SwitchAI Enabled

Key Highlights

Led global AI teams across multiple locations.
Developed advanced AI models for content intelligence.
Published research in top-tier AI conferences.

Stackforce AI infers this person is a leader in AI-driven solutions for social media and e-commerce industries.

Contact

Skills

Core Skills

Large Language Models (llm)Multimodal Learning

Other Skills

VLM2VecOmni modelVision Language ModelVision-Language Models (VLM)Multimodal AI systemsRecommendationdeduplication systemRecommender SystemsSearch EnginesData ScienceMachine LearningData MiningAlgorithmsPythonNatural Language Processing

About

Experience

9 yrs 8 mos

Total Experience

2 yrs 5 mos

Average Tenure

5 yrs 11 mos

Current Experience

Tiktok

4 roles

Engineering Manager - TikTok AI (Content Intelligence)

Promoted

Mar 2026 – Present · 2 mos

Leading a global AI research and engineering team across Singapore, Beijing, Shanghai, and San Jose. Our team focuses on building large-scale AI systems and high-performance models that power TikTok’s Content Intelligence.
Our work supports several critical downstream applications on the platform, including content diversity, content similarity detection, unoriginal content identification, deduplication, content safety, copyright protection, and the Creator Rewards Program.
From a technical perspective, we develop and deploy advanced LLM, VLM, and multimodal (“Omni”) models capable of understanding multiple modalities available on the platform, including text, images, video, and audio. We pre-train internal foundation models and adapt them for large-scale production use across a wide range of downstream applications.
In addition to product development, we actively collaborate with academic institutions such as the National University of Singapore and its high-performance computing labs on cutting-edge research directions. Some of our research areas include:
Long-video understanding and grounding
Efficient small language models
Advanced vision backbones and multimodal representation learning
Our team regularly publishes research in top-tier AI conferences and files patents based on our innovations.
We currently have multiple openings across all locations. If you are interested in working on high-impact AI model development at TikTok, please feel free to reach out to me.

Large Language Models (LLM)VLM2VecOmni modelVision Language ModelMultimodal Learning

Manager - TikTok AI (Content Intelligence)

Jan 2024 – Mar 2026 · 2 yrs 2 mos

Tech Lead (AI-Lab)

Promoted

Mar 2022 – Jan 2024 · 1 yr 10 mos

Senior Research Engineer (AI-Lab)

Jun 2020 – Mar 2022 · 1 yr 9 mos

Shopee

Senior Data Scientist

Mar 2019 – Aug 2020 · 1 yr 5 mos · Singapore

Built a detection and segmentation model for thousands of product types to incorporate in the image search (product discovery) on Shopee platform (Apps).
Built tools for AI-driven personalize (recommendation) auto Poster/Campaign/Banner generation to reduce the manual effort of the design team.
Built a generic facial recognition system for one-click registration and validation for millions of identities (users) with 99% accuracy.

Dataweave pvt. ltd.

Data Scientist (Data Semantics)

Feb 2017 – Jan 2019 · 1 yr 11 mos · Bengaluru Area, India

Leads end-to-end pipeline of multiple core data science projects.
1. Product Page Clustering: Developed deep learning model for image and text-based product page clustering to find similar products across thousands of e-commerce stores containing millions of products. This includes attribute and value normalization, attribute classification, similarity function definition, classification of products based on their features and designing a normalization layer based on various text-based attributes and features for every product. I am leading this product page clustering project.
2. Attribute Tagging: Worked on CRF based brand tagging project. This is an internal project use to tag brand to a given product description from thousands of e-commerce stores. Alongside Markov model, we also experimented with LSTM for tagging.
3. Sentiment Analysis and Product Review Summarization: Have wide knowledge on sentiment analysis and summarization of product reviews. Used word2vec and RNN for product review summarization.
4. Description Parsing: Developed Deep Learning model for product description parsing. In e-commerce platform description of a product is written in plain text which is a rich source of metadata for that product. This description is error-prone and too much noisy. Building deep learning model to parse this description to extract meaning full information and specification of the product.
core competencies in Computer Vision, ML, and Python.

Riverbed technology

Member Of Technical Staff

Jul 2016 – Feb 2017 · 7 mos · Bengaluru Area, India

Experience in: C++, Python, Hyper-v, Microsoft Azure.
Job profile: Developer at SteelFusion product team.
Worked on design of virtual granite core (SteelFusion for virtual platform) for Hyper-V.