Visualizing complex geoscience, wellbore, and seismic data through an intuitive, scalable, and AI-enabled architecture.
i2G is a high-performance, cloud-based SaaS platform engineered for geoscientists and operators in the energy sector. It acts as a nexus to visualize, interpret, and manage over 144,000 OSDU objects, from wellbore logs to seismic surveys.
As a Senior Product Designer & UI Engineer, my mandate was to overhaul an incredibly dense, legacy data structure into a modern, responsive interface. The goal wasn't just to make it look "clean"; it was to fundamentally change how professionals extract insights from massive, multi-gigabyte datasets over Minio storage.
The energy sector deals in massive, highly-technical tables (API numbers, Operator Details, Trace Data, Water Depths). Geoscientists were previously bombarded with disjointed grid views. The tension lay in balancing extreme informational depth without causing cognitive fatigue. Users needed geospatial context before they needed row-level data.
Traditional handoffs in zero-to-one enterprise builds are painfully slow. I needed to act as a multiplier—bringing over 7 years of deep product thinking to the table while leveraging modern AI-assisted engineering tools to bridge the gap from Figma prototypes to a live, Vercel-deployed React application in record time.
"The biggest challenge was translating a database-shaped platform into a workflow-shaped experience—and then literally coding it into reality without breaking pace."
I approach product design as an end-to-end discipline. Utilizing AI simply scales my execution velocity, allowing me to focus on architecture rather than boilerplate.
Defined the OSDU-compliant information architecture, establishing map-discovery patterns versus tabular constraints.
Used Figma Make for rapid structural explorations and layout validations, achieving pixel-perfect component libraries rapidly.
Stepped entirely out of Figma to scaffold the React application, writing live components and complex UI states using AI-assisted IDEs.
Shipped the fully responsive, functional UI directly to a live environment for client review. Zero translation loss from design to code.
To anchor users in geospatial reality before drowning them in data tables, I designed an interactive global map as the primary entry point into the platform. Clicking any well or seismic pin triggers a smooth, contextual right-hand sidebar — surfacing critical metadata (Platform ID, Water Depth, Operator, Target Status) without breaking spatial awareness.
The same robust design system and progressive disclosure pattern was deliberately applied to the Well Domain and Seismic Domain pages. Rather than reinventing the UI for each data type, I architected a single, scalable component library — clean data grids, pill-tag taxonomy, monospace ID typography — that generalizes cleanly across wellbore logs, seismic surveys, and rig data alike. One system to rule complexity across three distinct geoscience domains.
Analytical rigor demands extreme granularity. I engineered a progressive disclosure workflow moving from the macro to the micro layer. Clicking into a physical asset from the map triggers a focused metadata modal categorizing properties (General, Data, Permissions, Legal) to eliminate clutter. Digging deeper reveals lists of specific log data files. Instead of dumping raw numeric tables, selecting a log instantly renders full-screen, interactive graphical plots. Visualizing these dense technical measurements completely streamlines the interpretative process.
Rather than relying solely on complex filtering UIs, I integrated an "ASK AI BOT". Geoscientists can query the system natively ("Summarize metadata for Bravo-002"). The AI interface works as a pervasive floating action modal, instantly transforming the SaaS platform into an intelligent analytical partner.
Two of the most operationally critical surfaces in the platform: a global indexing search that lets users instantly find any of the platform's 144,000+ OSDU objects by keyword — eliminating dead-end navigation between domains — paired with an executive dashboard that consolidates the entire Data Space into a single command view.
The dashboard surfaces live KPIs: total object counts, active processing pipelines, Minio storage utilization metrics, and data ingestion health indicators. Designed specifically for operations managers and platform administrators who need a real-time pulse on the system without navigating individual domain pages. Two entirely different user archetypes, served by two purpose-built surfaces within the same cohesive design system.
A SaaS product is only as trustworthy as the control it gives users over their own experience. I designed a comprehensive Settings suite that proves i2G isn't just a slick front-end — it's a fully architected, production-grade enterprise platform built from the ground up.
The settings ecosystem includes: Profile Management for individual user identity, a Notifications Centre that surfaces real-time system events (data processing completions, maintenance schedules), Security controls including password management and two-factor authentication, a Session Manager for monitoring and revoking active logins across multiple devices, Data Management tools covering auto-backup scheduling, manual backup triggers, and bulk data import/export operations, an Appearance toggle for dark mode — respecting the shift-worker reality of geoscientists — and a System panel surfacing platform diagnostics.
This wasn't a checkbox exercise. Every settings page was designed with the same intentionality as the core product flows — clear hierarchy, contextual help text, and zero ambiguity. The result is a platform that feels complete and enterprise-ready to any operator who ever opens it.
The i2G redesign proved that complex enterprise layouts don’t have to look archival. By deeply integrating AI into my own design workflow, I was able to rapidly output high-fidelity systems, transition them directly to front-end development, and produce a live artifact that stakeholders could touch and feel instantly.
The end result is an intelligent, highly cohesive energy SaaS that feels lightweight and modern, while confidently supporting heavy OSDU schemas beneath the surface.
If I were to do this project again, I would spend even more time on the prompt architecture for the AI co-pilot integration. Teaching the AI to reliably retrieve specific geoscience metadata through a chat interface required significant iteration. It pushed my skills beyond just pixel-pushing into the realm of designing "conversational architecture."
This project ultimately codified my belief in the AI-Augmented Designer. With tools like Cursor and Claude Code, the limitations of traditional prototyping fade away. As a designer with engineering roots, being able to scaffold out my own components and immediately validate them in a live browser accelerates the feedback loop exponentially. The future of product design isn't just drawing interfaces; it's orchestrating them into reality.
"Jastej took our complex legacy platform and transformed it into something our team actually wants to use. The AI chat feature was a game-changer — our geoscientists can now query data naturally instead of navigating through endless menus. And seeing it live dynamically via Vercel during reviews completely changed our process."