Sohel Sura

After that, I started working on a much wider range of projects and gradually got comfortable working across different parts of the stack. Every project brought something new, whether it was application architecture, APIs, data, or infrastructure. This eventually led to long-term work with WP Provider and Kitkool, where I got the opportunity to work on larger and more involved systems.

At WP Provider, I worked on complex features for a hosting platform and started going deeper into AI and agent-based development. I explored areas such as RAG, vector search, and semantic matching, which gave me a better understanding of how AI can work with application data and existing software systems.

That interest eventually took me deeper into cloud infrastructure and DevOps. For Gamer Stats, I took care of the complete cloud setup alongside development, working with its large datasets and building AI and data-driven solutions around them. This included deployments, CI/CD, monitoring, and the infrastructure needed to keep the platform running reliably.

As AI tooling evolved, I started spending more time with AI coding agents and agentic development. I’ve worked with both commercial and open-source approaches, along with different models, and became interested in understanding how these systems actually work with a codebase. I’ve explored context management, tool use, MCPs, subagents, and how these agents can work within an existing software architecture.

I also put this into practice with Gamer Stats, where I built an MCP server that allows clients to connect the application to their AI environment and access its context and capabilities directly through an AI chat. This lets an AI agent work with the actual application and its available tools instead of treating it as a separate system.