AI / GIS Enterprise Interface
At SenseTime in Abu Dhabi, Amit Verma led graphics-heavy frontend engineering for AI-assisted enterprise applications built for ADGM. Vue, WebGL, Unity XR, and GIS sit in one interface, so teams can work with complex operational and spatial data inside real production constraints.
Problem
ADGM needed advanced machine learning to be usable inside enterprise interfaces that blend AI, GIS mapping, and Unity XR.
Approach
Build a Vue and WebGL frontend that carries AI output, spatial GIS data, and XR in one production interface.
Outcome
Enterprise tools where operational and spatial data is explored through interfaces that stay usable under production load.
Context
SenseTime's Abu Dhabi work serves ADGM, where the demands of a rigorous enterprise environment meet advanced machine learning. The brief asks for several things at once: surface AI capabilities, operational data, and spatial information inside web applications that internal teams and clients can actually use day to day. The hard part is rarely a single technology. It is making machine learning, realtime 3D, GIS, and XR behave like one product rather than a set of demos, and keeping that dependable under production load.
Challenge
AI outputs are probabilistic and often abstract, while enterprise users need interfaces they can read and rely on. The data is also both operational and spatial, which pulls in GIS for mapping and location context and Unity XR for immersive views, alongside the WebGL graphics work the web layer needs. Each of those stacks has its own runtime model and performance profile, and bringing them together without the interface becoming heavy, confusing, or fragile is the main engineering problem.
Approach
The architecture is built on Vue, so the application has a stable, component-driven backbone that the heavier graphics and AI features plug into. WebGL handles the realtime visual layer in the browser, GIS-based interfaces carry the spatial and mapping dimension, and Unity XR is integrated where an immersive view actually helps rather than for novelty. Machine learning is treated as a bounded input: the job is to frame model outputs so enterprise teams can read them at a glance and act on them. Delivery runs on containerised builds with Docker, CI/CD pipelines, automated testing, and production monitoring.
Engineering decisions
Keeping Vue as the foundation means the AI, WebGL, GIS, and XR pieces each sit behind clear boundaries instead of bleeding into one another, which is what keeps a multi-discipline interface maintainable. WebGL and Unity XR are used for the parts that genuinely benefit from realtime 3D and immersion; standard enterprise UI patterns carry everything else, which protects both performance and comprehension. That restraint comes from the earlier browser-graphics and games work, where interaction quality and runtime performance were most of the job. The rule I work to here: bring in the heavy capability only where it makes the product faster to use.
What it demonstrates
Senior frontend and WebGL engineering applied to the AI and enterprise market, in the Abu Dhabi context specifically: owning the interface side of AI-assisted products end to end, combining Vue, WebGL, Unity XR, and GIS into one coherent application, and shipping it against microservice backends with Docker, CI/CD, and automated testing.
Outcomes
- Led graphics-heavy AI-assisted enterprise interfaces for ADGM, owning the frontend architecture on Vue.
- Integrated Unity XR experiences and GIS-based interfaces into internal and client-facing platforms.
- Framed machine learning output inside product design systems that stay usable under production constraints.
- Shipped against microservice backends with containerised builds, CI/CD pipelines, automated testing, and production monitoring.