Developing a domain-specific CRM based on EspoCRM to address the specialized workflows, terminology, and operational requirements of freight forwarding companies.

The project involves designing custom entities, relationships, workflows, and business processes, translating real-world business requirements into scalable CRM functionality. The platform is continuously evolving toward a flexible CRM solution purpose-built for freight forwarding operations.

Developing an AI-powered system that makes large Outlook email archives accessible through natural-language interaction.

Using a Model Context Protocol (MCP)-based architecture, the system connects to locally stored Outlook .ost data and exposes structured email context to AI models. It enables natural-language search, filtering, conversation exploration, cross-email analysis, and automated summarization.

The project focuses on transforming unstructured email archives into AI-accessible context, while continuously improving retrieval accuracy, contextual relevance, and the quality of generated insights.

Built an interactive Apache Superset BI dashboard backed by a PostgreSQL data warehouse, supported by automated Python/Pandas ETL pipelines for data processing and reporting.

Implemented Row-Level Security and role-based access controls, enabling secure multidimensional analysis across key business dimensions. The solution reduced manual reporting effort by approximately 5 hours per week while improving access to actionable business insights.

Developed a lightweight Python/Pandas utility for standardized data cleaning and rapid data-quality profiling. Built a chainable df.clean API supporting common preprocessing tasks such as column normalization, missing-value handling, deduplication, datetime normalization, and categorical cleanup.

Added automated column-level profiling and robust handling of messy real-world datasets, creating a reusable abstraction that reduces repetitive data-preparation code and accelerates exploratory data analysis.