AI systems · research software · scientific computing · computer vision · reproducible engineering

I build systems that sit in the messy middle between research ideas and usable engineering.

My work usually starts with an experimental problem: a dataset that is hard to trust, a model that works only in someone’s notebook, a manual workflow that slows down research, or a pipeline that breaks the moment it leaves the original machine. I like turning those pieces into something cleaner: reproducible, testable, documented, and usable by people who did not build it.

I am currently studying Electrical and Computer Engineering at Vanderbilt University and working across AI research engineering, computer vision, scientific image analysis, and applied machine learning systems.

I am most interested in AI that becomes useful outside the demo.

That means I care about:

• models that can be evaluated honestly, not just shown beautifully
• research code that someone else can actually run
• pipelines that preserve assumptions, outputs, and failure cases
• scientific tools that reduce manual work without hiding uncertainty
• systems where the interface, backend, model, and validation all matter

I do not think of software as separate from research. For me, software is often the instrument that makes the research possible.

My technical path has moved through several connected areas:

• AI-assisted materials research and AFM image analysis
• computer vision for classification, segmentation, and detection
• GPU-accelerated model training and inference
• scientific workflow automation
• reproducible research platforms
• data engineering for multilingual AI systems
• simulation-guided mechanical design and CAD workflows
• computational chemistry and molecular modeling

Across these, the pattern is the same: understand the scientific or engineering problem first, then build the system around it.

I am building toward research engineering roles where AI, scientific computing, and production systems meet.

The kind of work I want to do long-term:

• build AI tools for scientists and engineers
• design reliable ML pipelines for real-world data
• work on computer vision and multimodal scientific analysis
• contribute to open-source research infrastructure
• eventually pursue graduate research in AI, ECE, robotics, or scientific machine learning

I am especially drawn to problems where the model is only one part of the system. The real challenge is making the whole workflow trustworthy.

I like direct feedback, measurable progress, and systems that survive contact with reality.

My default engineering instincts:

• make it reproducible
• make it understandable
• test the boring failure cases
• document the assumptions
• avoid pretending uncertainty does not exist
• build for the next person who has to use it

I value clean architecture, but I value working research systems more. A good system should not only run once. It should be explainable, recoverable, and useful after the first exciting result.

I came into engineering through curiosity, robotics, computational chemistry, and a long habit of trying to understand how complex things behave underneath the surface.

Over time, that turned into a stronger focus: building tools that help people analyze, verify, and accelerate difficult technical work.

I am ambitious, but not because I want a polished résumé. I want to build things that make hard research easier to do, easier to reproduce, and easier to trust.

• GitHub: NIne-WIngEd
• LinkedIn: MK Rayan
• Email: md.mostakim.khan@vanderbilt.edu