Yug Sinha

2+ years of work experience

Jan 20268 mosPresent

Bangalore Urban

- Worked on Voice AI Stylist with sub-second responses, handling all edge cases and user queries to handle users at scale. – Architected low-latency data layer across MongoDB, Redis, Qdrant, and APIs, powering fast image-based semantic product retrieval and personalized ranking – Implemented FalkorDB (Redis based Graph VectorDB) agent memory pipeline with <200ms writes and <10ms reads, enabling real-time context persistence and ultra-fast conversational recall. – Architected retail assistance, a live powered agentic shopping assistant with intent routed tool capabilities and structured widget types published over a real-time data channel – Average response time being 1s along with selectable widget publishing capability, deeplink navigation across iOS, Android, Mobilesite clients – Architected multistore persistence layer, MongoDB for session snapshots and chat history, Qdrant for vector search, and in memory state store achieving sub 100ms turn latency for all non-LLM operations.

Jul 20257 mosJan 2026

Bangalore Urban

- Built a Load Balancer for rerouting multiple databricks user requests into multiple workspaces to increase DAU handling at peak traffic times – Worked on Langchain, AzureOpenAI Server, implemented Azure AI search for Unstructured Data, worked on Txt to SQL model for Structured Data retrievel for client specific solutions. – Created an end to end to RAG Pipeline in databricks to pull from a digital library of over 1TB of documents and to ingest in a Datastax vector tables for RAG Chatbot.

Dec 20247 mosJun 2025

Bengaluru, Karnataka, India

Aug 20235 mosDec 2023

Fine-tuned models for industry-specific use cases, reducing bias and errors, and deploying large-scale Al models. – Collaborated with cross-functional teams, incorporating user feedback into platform improvements.