从数据到智能：企业级数据平台的构建、演进与 Agentic BI 实践。一个数据开发的八年数据工程手记：从 0 到 1 构建企业级医药数据平台，再到 Data + AI 转型下的 Agentic BI。

Code for the paper "Match, Compare, or Select? An Investigation of Large Language Models for Entity Matching" (COLING 2025)

The official implementation of [Quality over Quantity: Boosting Data Efficiency Through Ensembled Multimodal Data Curation] in AAAI2025.

Our project for the "Data Intelligence Applications" exam at Politecnico di Milano. The project was about Social Influence and Pricing techniques applied to networks.

Decision intelligence platform for industrial manufacturing. Connects to CRM, ERP, and CMMS systems and monitors industry and macroeconomic conditions to compute leading indicators, generate predictions, and deliver daily executive briefings.

MCP integration to use Cassis with Snowflake when Cassis isn't natively connected to the warehouse. Cassis generates grounded SQL, Snowflake executes it.

A structured intelligence layer for internet attention.

PM observability KPI framework for a real-time clinical risk-stratification stack (Mobile SDK → APIM → Kafka → ETL → XGBoost → 3rd-party).

Patent intelligence for AI agents — patent search, USPTO data, patent landscape & pgvector prior-art search. MCP + x402.

Weather & climate intelligence for AI agents — current weather, forecast, historical, climate normals, alerts, agricultural & travel weather. MCP + x402.

Trade-Based Money Laundering investigation reports generated in minutes — not weeks. This actor is built for AML compliance officers, trade finance banks, and financial investigators who need a scored TBML risk assessment backed by data from 14+ authoritative sources and five independent forensic algorithms.

Gati Shakti AI Digital Twin is an AI-powered governance intelligence platform designed to revolutionize infrastructure planning and execution in India. It acts as a cognitive digital twin of national infrastructure, enabling real-time coordination across ministries, predictive decision-making, and citizen-driven governance.

Cyber attack surface report for any domain — enter a target and get a full external risk assessment in minutes.

Data visualizations through Tableau for insightful analytics and decision-making using Walmart retail data.

AI-powered job screening system that helps match candidates with job openings

Batch data intelligence platform. Natural language queries → structured, cited datasets. Deployed as an MCP server.

Django-based job-search analytics platform with funnel metrics, data quality checks, workbook exports, evidence documentation, and 133 passing tests.

Open Source Intelligence MCP — repo health, dependency risk, trending, license checks. FoundryNet Data Network.

The Cognitive Node of the Automated Data Intelligence Platform (ADIP). An AI-powered analytical infrastructure that consumes raw data from the Ingestion Engine into automated insights, forecasts and applied LLM reasoning, all served via Streamlit.

DomainKits connects Claude to live domain data. Built-in workflows help verify domains from multiple angles before making decisions. Supports domain search, analysis, brand conflict detection, valuation, trend discovery,