Building the platform training India's next 500 AI engineers

When we started, Lex AI had a website, but it wasn't doing the job. The logo wasn't right. There were no brand guidelines, no design system, no visual language that could carry across a website, a learning platform, and university-facing print materials. The foundation had to be rebuilt.

We started where it mattered most: identity. Who is Lex AI? What does it look like? What does it feel like when a university dean opens a brochure or a student lands on the website?

We designed 34 logo variations. Not random iterations - each one explored a different angle of what Lex AI could mean visually. Puru picked the one that felt right, and we built the entire design system around it. Color palette, typography, spacing, component library. One design language that would carry across every touchpoint.

Then we rebuilt the website. lexailabs.com went from concept to deployed in under a week. But launching fast didn't mean launching rough. The site had to convince engineers that this was a serious learning platform, convince university deans that this was a credible institutional partner, and convince enterprise leaders that Lex AI understood boardroom-level AI strategy. Three audiences, one website, one cohesive story.

Prof is Lex AI's core learning platform for individual engineers. AI and ML taught the way Karpathy and Hassabis learned it. Not tutorial walkthroughs. Not API wrappers dressed up as courses. First-principles engineering with real projects, a Socratic AI tutor that asks questions instead of giving answers, and interview prep mapped to actual rubrics from Google, Meta, OpenAI, and Anthropic.

We built Prof end-to-end. Frontend in Next.js. Backend on Google Cloud Run. Content engine in Docusaurus for the course material. Razorpay for payments. Google authentication for student access. Admin dashboards for Puru's team to track enrollments, progress, and revenue.

The platform went through multiple iterations before we arrived at the architecture that worked. That's how real product development goes. You don't get it right on the first try. You build, learn what breaks, understand what the business actually needs - not what it thinks it needs - and rebuild until the system fits the problem perfectly.

We shipped it with monitoring, deployment pipelines, and infrastructure built to handle real usage without breaking.

Lex AI wasn't just a platform. It was an ecosystem being built in real time. Puru was getting invited to speak at IISc, GITEX Dubai, EduTech Asia Singapore, and university campuses across India. Every event needed collateral.

We designed and delivered:

• →An "AI for Leaders" brochure for enterprise meetings
• →ML book covers for Puru's textbook on machine learning fundamentals
• →University pitch brochures for critical meetings
• →Business cards
• →Demo videos of the Prof platform in action

Each of these sounds small in isolation. But when you're building a company's entire visual and digital presence, everything compounds. The brochure that goes to a university dean uses the same design language as the website that a Google engineer sees when they search for AI courses. That consistency is what makes a young company feel like it's been running for a decade.

Most agencies hand you a Figma file and a “good luck.” We handed Lex AI a system that makes sure everything built after us looks, feels, and works like everything we built.

The design system wasn't a PDF. It was a living reference - exact color tokens, typography scale, spacing rules, component patterns, and usage guidelines. Whether it's a new landing page, a new course module, or a conference banner, the system answers every visual question before it's asked.

We structured everything so that any developer using AI coding tools like Cursor or Claude Code can maintain brand consistency without asking anyone. Lex AI's visual consistency doesn't depend on one designer being available. It's built into how the codebase is documented.

The handover was thorough enough that Puru's next team could pick up exactly where we left off without a single phone call.

We didn't build a website and walk away. We didn't scope a platform and hand off a spec. We became part of the company for five months. We understood the business well enough to design a university brochure on short notice because we already knew who the audience was, what the product did, and why it mattered.

And when the engagement ended, we didn't leave a gap. We left a design system that enforces itself, documentation structured for AI-native development, and knowledge transfer thorough enough to make the next team productive from day one.