AI/ML Engineer
Clicode/Xappiens
Generative AI applications, agentic workflows, document intelligence, and enterprise assistants. Integrating LLM providers and optimizing retrieval for accuracy, latency, and cost.
I'm Dhrumil Prajapati. I build AI systems that connect knowledge, conversations, and decisions — from RAG pipelines to voice assistants.
Engineering useful AI, with curiosity for everything around it.
I'm Dhrumil, an AI/ML engineer based in Ahmedabad, India. I build practical AI systems that bring together knowledge retrieval, agent workflows, and conversational interfaces.
At Clicode/Xappiens, my work spans generative AI applications, document intelligence, and enterprise assistants. Earlier, I developed RAG pipelines at Tuvoc and researched satellite-image change detection at ISRO's Space Applications Centre.
I also teach computer skills. Making a complex idea understandable is part of how I approach engineering, too.
Explore the stack, from agent reasoning to production delivery.
Give reasoning a structure.
From Earth observation research to applied AI product engineering.
Generative AI applications, agentic workflows, document intelligence, and enterprise assistants. Integrating LLM providers and optimizing retrieval for accuracy, latency, and cost.
Built a Whisper-to-retrieval pipeline covering transcription, preprocessing, embeddings, and vector indexing. Benchmarked OpenAI, Instructor, and MiniLM embeddings and tuned top-k and distance metrics.
Developed a U-Net / ResNet-34 change-detection pipeline using OSCD and custom-labeled Sentinel-2 imagery of Ahmedabad. Prepared GeoTIFF inputs and generated change masks.
Good software starts with a clear problem. AI coding tools help turn that intent into working code—from exploring an unfamiliar codebase to iterating on an implementation.
My approach: give the agent useful context, work in small changes, and review the result. Architecture, testing, and the final decision stay part of the engineering process.
Hover or tap to explore the tools.
Four projects, four different problems. Explore the code and the ideas behind each one.
An AI project focused on the job-search space. Explore the repository for implementation details and current progress.
Turn meeting audio and transcripts into summaries, action items, and searchable knowledge. A meeting intelligence project with semantic retrieval.
An AI assistant project exploring multimodal interaction. Open the repository to follow the code and current development.
An AI trip-planning project. Explore the Python implementation and repository for the current planning workflow.
Government Engineering College, Modasa
Gujarat Technological University
View full image ↗
View full image ↗
View full image ↗
View full image ↗Let's talk about AI engineering, practical products, or teaching.
[email protected] ↗