Software Engineer (AI / Agentic Systems) in Penarth

Synscribe is building the next-generation SEO agency as a service — not tools for consultants, but AI systems that replace the consultant layer entirely. We are hiring a software engineer to help us design and build AI agents that:

• Plan, execute, and iterate on SEO strategies autonomously
• Learn from outcomes over time
• Deliver measurable revenue impact for high-growth B2B startups

This is not a traditional SaaS role. This is "SEO tooling." This is applied AI, agents, evals, and real-world feedback loops, deployed directly into customer production environments.

About Synscribe: Synscribe is a fast-growing B2B service company at the forefront of AI-driven content marketing and AI search (LLMs / GEO). We partner with Series A–B B2B SaaS companies (typically $500k+ ARR, global) to ensure their ideal customers find them across:

• Traditional search engines (Google)
• AI-native search and discovery surfaces (ChatGPT, etc.)

What makes us different:

• Results in days, not months. Clients often see inbound leads within 3 days of publishing.
• LLM-native impact: AI search regularly drives ~2× the leads of traditional search, 2–4× higher conversion rates, and 10× operational efficiency vs traditional SEO agencies.
• Engineering-led agency model: We build and operate proprietary AI systems instead of hiring armies of consultants.

What We’ve Built So Far: You’ll be working on (and extending) a production system that already includes:

• AI-powered keyword & content planning
• Deep competitor research
• Social listening → content ideation
• Automated, high-quality content pipelines
• Multi-step research
• Citation + FAQ generation
• Structured outputs
• Programmatic landing page generation
• Automated link building & outreach
• End-to-end planning → execution → reporting loops

These systems are used daily to drive revenue for real companies.

What You’ll Work On:

1. The Next-Gen "SEO Agency as a Service": We are building AI systems that replace humans, not just assist them. That means:
• Encoding the skills of the best SEO consultants into scalable AI agents
• Planning and executing work over weeks and months, not single prompts
• Learning from past outcomes to improve future strategies
• Owning customer outcomes, not just providing tools
2. Forward-Deployed Engineering: You’ll work directly with startup founders, in-house engineers, and growth teams. You won’t just ship features — you’ll:
• Implement changes directly in customer environments
• Close feedback loops fast
• Discover real constraints early
• Use production data to inform system design
3. Hard, Interesting Problems We’re Actively Solving:
• Output quality at scale - LLM evals, human-in-the-loop systems that don’t overwhelm humans
• Formalized agent "skills" - Long-horizon planning
• Tool use
• Self-improvement from historical results
• AI model collapse / ouroboros problems - Avoiding AI consuming AI-generated trash
• Designing robust RAG and core knowledge boundaries
• Agency → self-service transition - Lowering the education barrier
• Making complex systems usable without SEO expertise
• Ecosystem-level effects - Cross-portfolio link building
• Network effects across customers
• Human-as-trainer loops - Systems that learn from expert operator behavior
• Continuous improvement without prompt spaghetti

Tech Stack:

• TypeScript (real TypeScript, not any)
• Next.js
• Trigger.dev (durable execution / workflows)
• LangChain
• LangGraph / Mastra
• Internal orchestration layer for agent + human-in-the-loop workflows

Advanced Topics You’ll Touch:

• AI agents & agent orchestration
• LLM evals and quality measurement
• RAG and structured knowledge systems
• Human-in-the-loop system design
• Longitudinal learning and feedback loops

Culture & How We Work: This is a culture-defining hire. We are a small team (currently 3) moving very fast. You will help set the bar.

What we value:

• High ownership: You care deeply about your work, your craft, and the customer.
• Independent learning: You don’t wait to be taught. You bring net-new knowledge into the team.
• Experimentation over ego: Hypothesize → test → validate → iterate. Ideas don’t win — results do.
• High-leverage work: Do the dirty work, then automate it. If you’re stuck doing low-leverage work repeatedly, something is broken.
• AI-first (but not AI-dependent): AI is an exoskeleton, not a crutch. You understand its limitations as well as its power.

Must-Have Skills (No Exceptions): If you don’t meet all of these, this role is not a fit.

• Experience with AI coding tools (Claude Code preferred; Cursor / Codex acceptable)
• Strong TypeScript
• Next.js experience including server-side execution, constraints, and tradeoffs
• At least one real project using LLMs + structured data
• Craftsmanship: You care deeply about the quality of what you build and continuously push yourself to improve.

Nice-to-Haves (Stand Out Here):

• Blogs or write-ups on things you’ve built (especially AI-related)
• Open-source projects
• Prompt engineering experience
• LLM evals or quality measurement experience

Who Will Not Thrive Here: This role is not for you if:

• You need hand-holding or constant direction
• You haven’t learned or tried anything new recently
• You consume content about AI more than you experiment
• You can’t work without AI tools
• You don’t understand the current limitations of AI
• You’ve been doing the same thing month after month
• You believe you’re "too senior" for unsexy or dirty work

Level & Structure: No management roles, no PM buffer, high autonomy, high responsibility.

Why This Role Is Special: You won’t be maintaining legacy code, tweaking dashboards, or building features disconnected from outcomes. You will be designing real AI systems that operate in the wild, shipping fast and learning faster, and helping define what an "AI-native agency" actually looks like.

If this excites you, we’d love to see what you’ve built.

Where to apply? https://www.synscribe.com/career#J-18808-Ljbffr