Mikel K

AI MCP Engineer

CData Software • Temps plein

Sep 2025 - Present • 1 yr 2 mos

- Leading R&D on Model Context Protocol (MCP) servers, focusing on performance, scalability, and real-world usability across LLM ecosystems - Designed and optimized a custom wrapper for the CData MCP Server, improving flexibility, integration, and efficiency - Conducted a 3-month comparative analysis of CData vs Monday Native MCP Servers, delivering actionable optimization insights - Reduced token consumption and response payload size while maintaining high accuracy and clarity - Built and refined MCP tools, prompts, and workflows for consistent performance across environments (e.g., Claude Desktop) - Developed LangChain-based agent workflows with custom logging, token tracking, and performance monitoring - Created analytics tools with a custom UI to visualize token usage, latency, and response efficiency - Implemented a RAG system using vectorized metadata with hybrid retrieval (cosine similarity + BM25) - Integrated OpenAI embeddings to enhance semantic search and retrieval quality - Improved system performance and response times through continuous benchmarking and optimization - Fine-tuned Llama and Qwen models for metadata retrieval; built an MCP-integrated pipeline combining programmatic fetching, RAG, and local LLM inference for high-accuracy context generation

Technical Engineer

CData Software Inc. • Temps plein

Oct 2024 - Sep 2025 • 11 mos

Technical Engineer with hands-on experience supporting and extending CData’s data connectivity platform. I work directly with customers and internal teams to troubleshoot complex integration issues across databases, SaaS applications, BI tools, and ETL pipelines. In addition to support and solution design, I actively work with the CData codebase, analyzing driver behavior, debugging protocol-level issues, and contributing fixes and enhancements when needed. My role frequently involves deep dives into SQL engines, ODBC/JDBC drivers, REST and SOAP APIs, authentication mechanisms, and performance optimization. I collaborate closely with engineering to reproduce issues, validate fixes, and improve product reliability, while also translating complex technical concepts into clear guidance for customers. Comfortable working across cloud, on-prem, and hybrid environments, and experienced with enterprise systems, security constraints, and real-world data integration edge cases.