Sandipan Das

About Sandipan Das

I am an AI and Data Science professional passionate about solving real-world problems through cutting-edge technologies in language modeling, multimodal learning, and traditional machine learning. With experience across Natural Language Processing (NLP), Computer Vision, and ETL workflows, I specialize in building end-to-end pipelines that extract insights, automate decisions, and enhance human understanding.Currently, I’m developing advanced Multimodal Retrieval-Augmented Generation (RAG) systems that fuse text, images, and documents for enterprise-level semantic search and intelligent Q&A, optimizing performance through vector databases, FAISS indexing, and LLM integration. My work bridges structured and unstructured data—applying transformers, graph-based reasoning, and keyword-aware routing to drive business automation.Previously, I built dashboards, automated reporting systems, and scalable data pipelines using Python, Amazon QuickSight, and cloud-native tools to support strategic decisions for cross-functional teams. I hold an M.S. in Computational Data Analytics (Georgia Tech, ongoing), and a Postgraduate Diploma in AI from Delhi, where I explored deep learning architectures, including CNNs, RNNs, and probabilistic models like GMMs and HMMs.My expertise spans:Language Models (LLMs): GPT-family, BERT, LLaMA, fine-tuning with LoRA & QLoRAMultimodal AI: Vision-language models, image captioning, VQA, and joint embeddingRAG Systems: LangChain, FAISS, GraphRAG, hybrid retrieval, OpenAI/Together/Groq APIsComputer Vision: Object detection, image generation with diffusion models, explainable visionTraditional ML: XGBoost, clustering, time series forecasting, A/B testingBig Data & ETL: Spark, SQL, Python, Kafka, Oracle/PostgreSQL pipelinesDriven by curiosity and impact, I aim to build AI agents that reason, explain, and augment human capabilities in domains like legal tech, finance, and enterprise security.

Experience

1. Data Scientist (Ai Engineering)

   Cisco

   Feb 2024 — Present · Bengaluru, IN

   Designed and deployed Agentic AI workflows for real-time root-cause analysis and data lineage using LangGraph, LangChain, and OpenAI/GPT, enhancing data traceability and audit readiness by ~40%. • Implemented semantic search and topic clustering algorithms for unstructured data (emails, logs), improving search relevance and discovery rate by 30% across internal data portals. • Developed machine learning pipelines for classification and anomaly detection using XGBoost, LSTM, and ensemble models, driving business use cases like fraud detection, subscription churn prediction, and cost optimization, impacting $1.5M+ in business decisions. • Led the integration of generative BI capabilities into DataHub, enabling natural language querying, insight summarization, and automated report generation, reducing manual reporting efforts by 60%. • Collaborated with cross-functional BI and product teams to define KPIs and build self-serve dashboards (Tableau/Power BI) used by 100+ stakeholders, translating complex ML outputs into business-impact narratives. • Pioneered RAG (Retrieval-Augmented Generation) architectures for BI, enabling semantic analytics over relational data and improving decision intelligence in areas like supply chain, sales, legal teams and finance.

Education

• Georgia Institute of Technology

  Master of Science - MS, Computational Data Analytics

• Indraprastha Institute of Information Technology, Delhi

  Post Graduate, Data Science and Artificial Intelligence