AI-Native Web Developer

This is a remote position.

Design, build, and ship marketing sites and landing pages across the Spektra portfolio — CloudLabs, CSP Control Center, SaaSify, CloudEvents, and spektrasystems.com — with AI agents as your execution engine. Brand discipline,design,SEO, and craft, not just typing.

About the roleSpektra Systems runs four product brands (CloudLabs, CSP Control Center, SaaSify, CloudEvents) plus the parent company site at spektrasystems.com. Each one has its own audience, voice, and pricing motion — labs platform for ISVs and educators, CSP control plane for cloud distributors, SaaS-platform tooling for software vendors, events platform for hackathon hosts.You own the marketing surface across all of them. Most execution happens through AI coding agents — Claude Code, Cursor, GPT,whateverworks. What you bring isn’t typing. It’s taste, brand discipline, and the architectural judgment to know when to trust the agent, when to push back, and when to throw it out and start over.This isn’t a “prompt engineer” job. It isn’t a “frontend dev who uses Copilot sometimes” job either. It’s a hybrid: you direct AI agents to do the bulk of implementation, you review ruthlessly, and you own the outcome — across five distinct brands.

What you’ll do

• Ship marketing pages, landing pages, and microsites across all five Spektra properties using Claude Code (or equivalent) as your primary execution engine. You iterate via screenshots, browser inspection, and tight critique loops — not line-by-line typing.
• Hold each brand’s voice and visual identity distinct — CloudLabs is opinionated and transparent (publishes pricing competitors hide); CSP Control Center is procurement-focused and technical; SaaSify is partner-channel-led; CloudEvents is energetic and event-driven; Spektra Systems is the trust-and-compliance parent. You can rewrite any AI-generated copy so it sounds like the right one.
• Maintain design discipline across the portfolio — semantic tokens (no raw hex), component reuse where it fits, custom where it doesn’t. Catch the agent when it reaches forbg-emerald-500instead ofbg-primary.
• Own SEO end-to-end across every site: canonical URLs, redirects, structured data, internal linking, top-traffic-URL protection through migrations. You spot when the agent silently breaks SEO and you recover before it ships.
• Architect information for messy real-world products. CloudLabs alone sells four distinct things (pre-built labs, VM Labs platform, annual Platform contract, hackathons) with four different buyers. CSP and SaaSify each have their own audience archetypes. You design IA for that.
• Run competitive research and translate findings into design and content choices that fit each brand — not just bullet points.
• Decide when to share components across brands and when to keep them separate. Build a shared design-system foundation where it accelerates work, fork where each brand’s identity demands it.
• Push back on AI output when it’s wrong. The agent will produce visually generic, brand-violating, SEO-broken, or copy-bloated work confidently. Your job is to catch it before merge.

Requirements

What you’ll do

• Ship marketing pages, landing pages, and microsites across all five Spektra properties using Claude Code (or equivalent) as your primary execution engine. You iterate via screenshots, browser inspection, and tight critique loops — not line-by-line typing.
• Hold each brand’s voice and visual identity distinct — CloudLabs is opinionated and transparent (publishes pricing competitors hide); CSP Control Center is procurement-focused and technical; SaaSify is partner-channel-led; CloudEvents is energetic and event-driven; Spektra Systems is the trust-and-compliance parent. You can rewrite any AI-generated copy so it sounds like the right one.
• Maintain design discipline across the portfolio — semantic tokens (no raw hex), component reuse where it fits, custom where it doesn’t. Catch the agent when it reaches forbg-emerald-500instead ofbg-primary.
• Own SEO end-to-end across every site: canonical URLs, redirects, structured data, internal linking, top-traffic-URL protection through migrations. You spot when the agent silently breaks SEO and you recover before it ships.
• Architect information for messy real-world products. CloudLabs alone sells four distinct things (pre-built labs, VM Labs platform, annual Platform contract, hackathons) with four different buyers. CSP and SaaSify each have their own audience archetypes. You design IA for that.
• Run competitive research and translate findings into design and content choices that fit each brand — not just bullet points.
• Decide when to share components across brands and when to keep them separate. Build a shared design-system foundation where it accelerates work, fork where each brand’s identity demands it.
• Push back on AI output when it’s wrong. The agent will produce visually generic, brand-violating, SEO-broken, or copy-bloated work confidently. Your job is to catch it before merge.

What we’re looking for

• AI fluency. You’ve shipped real production work with Claude Code, Cursor, or similar. You know which model to reach for, when to start fresh vs. continue a session, when to spawn a sub-agent, and how to give an agent the context it actually needs (paths, line numbers, prior-art file references — not vibes).
• Design taste. You know that monochromatic isn’t always bad and that color variety isn’t always good. You don’t confuse“looks like a generic SaaS template”with“shipped.”
• Multi-brand judgment. You can hold five distinct voices in your head and switch fluently between them. You don’t homogenize everything to the same template.
• Frontend chops. Astro, Next, React, Tailwind v4, modern build tooling, OKLCH color, dark/light theming. You read code fluently even when you didn’t type it. You knowgit.
• SEO depth. You know what a canonical loop is and why it hurts. You know when a 301 is right and when it isn’t. You’ve broken SEO and recovered from it
• Bias to ship. You’d rather merge an iteration today than perfect tomorrow.
• Comfortable with brutal feedback loops. Most of what you build will get critiqued and reworked. You don’t take it personally; you take it as signal.

What we don’t care about

• Years of experience. Show us output.
• Whether you “can also code without AI.” Obviously you can. That’s not the point.

How to applySend links to two or three marketing sites, landing pages, or component libraries you’ve shipped — ideally ones an AI agent helped you build.In three sentences, tell us:

1. Whatyoudid vs. what the agent did.
2. The single dumbest thing the agent tried to ship that you caught.
3. What you’d do differently next time.

Originally posted on Himalayas