AI fluency with Claude Code, Cursor, or similar; shipped real production work with AI agents and know when to start fresh, spawn sub-agents, and provide necessary context.
Design taste and brand discipline; ability to hold five distinct brand voices and prevent homogenization.
Frontend chops: experience with Astro, Next, React, Tailwind v4, OKLCH color, dark/light theming; proficient in reading code and using Git.
SEO depth: knowledge of canonical loops, 301 redirects, structured data, internal linking, migrations, and recovering from SEO issues caused by AI.
Requirements:
Ship marketing pages, landing pages, and microsites across all Spektra properties using Claude Code (or equivalent) as the primary execution engine; iterate via screenshots, browser inspection, and tight critique loops rather than line by line typing.
Maintain distinct brand voices and visual identities across CloudLabs, CSP Control Center, SaaSify, CloudEvents, and Spektra Systems; enforce design system discipline and catch AI output that is brand violating or SEO broken.
Own SEO end to end across every site: canonical URLs, redirects, structured data, internal linking, and migrations; detect and recover SEO issues before shipping.
Architect information architecture for messy real world products and multiple brands; run competitive research, translate findings into design and content choices, decide when to share components across brands, and build or fork a shared design system as needed; push back on AI output when wrong.
Job description
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 role
Spektra 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, whatever works. 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 for bg-emerald-500 instead of bg-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 for bg-emerald-500 instead of bg-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 know git.
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 apply
Send 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:
What you did vs. what the agent did.
The single dumbest thing the agent tried to ship that you caught.
