Danish Anjum

Developed an Extended Reality (XR) smart glass application using Deep Learning and Computer Vision to detect and estimate vehicle damages in real time. Implemented OCR and Named Entity Recognition (NER) pipelines to extract structured information from Driver Licenses, License Plates, Insurance documents, and Odometer readings. Developing a mortgage document intelligence pipeline to classify 300+ document types from large multi-document PDF bundles using TF-IDF based NLP features and ML models (Random Forest, XGBoost, SVM, Naive Bayes) achieving ~85% accuracy. Implementing document boundary detection using a modified Hidden Markov Model (HMM) (~84% accuracy) and integrating GenAI capabilities with a RAG architecture (LLM embeddings, vector search, AI agents) for semantic document retrieval and natural language querying.

Client: Global Leader in Spirits & Wine (US). Consolidated multi-source datasets (~16K SKUs) with a master mapping framework, enhancing brand-level insights and supply chain tracking. Automated forecasting workflows and Azure pipelines (Databricks, ADF), reducing report generation time by 85% and enabling real-time KPI visibility. Developed a GenAI chatbot using Azure OpenAI & Django to enable natural language access to sales and market data, cutting insight time from days to minutes. Enhanced chatbot with RAG and Agentic RAG for improved multi-turn query resolution and contextual understanding. Achievements: Won 1st prize for developing a data-driven CLTV optimization solution during LatentView's POC engagement with ezCater.

Designed and built and education portal using HTML, CSS and JavaScript