Senior Founding Engineer – AI Learning Platform

Job description

Location: Hybrid (UK preferred) Reporting to: Head of Engineering Key Partners: Kim Faura (Product Lead) & Pravin Paratey (Head of Engineering) The Split: 80% Deep Building & Coding | 20% Technical Leadership & Team Shielding ABOUT SALESAPE & ABI We are building what we believe will become the operating system for millions of small businesses. Today, we have one main product brand — SalesApe, which helps businesses automate customer conversations, qualify incoming leads, and convert more sales. Alongside this, we are building Self-Serve Abi (Artificial Business Intelligence) — a natural language AI business partner that allows business owners to create, operate, and grow their businesses simply by talking to an AI. But our long-term vision goes far beyond individual AI agents. We believe the next generation of software will continuously learn from the outcomes it creates. Every customer interaction, recommendation, experiment, and business outcome should make the platform smarter for the next customer. To achieve that, we're looking for a Senior Founding Engineer to build the core intellectual property that ties these products together: our unified, self-improving Learning Platform. THE MISSION Your mission is to architect and build the intelligence layer that sits behind both SalesApe and Self-Serve Abi. This platform will capture business events, measure outcomes, identify patterns, and continuously improve the recommendations our AI makes. Rather than simply orchestrating existing foundational models, you will build a self-improving recommendation and learning engine that compounds over time. Imagine millions of businesses collectively teaching the platform: which sales techniques convert best, which marketing campaigns actually work, and which onboarding journeys reduce churn. Every customer benefits from the learnings generated by every other customer, strictly preserving privacy and security. This is not a theoretical academic exercise. We will frame your first phase deliberately: prove the learning loop end-to-end on one real, high-value problem—onboarding and retention—using the outcome data we are already generating, from our early trust and engagement signals to the events pipeline and retention dashboard. A concrete early win on a problem we genuinely care about earns us the right, and the real-world data, to tackle the harder architectural questions. From there, you scale and compound the same loop across the rest of the product. This approach drives immediate product value while we build toward the multi-year strategic defensibility moat we need ahead of our Series B. WHAT THIS ROLE ACTUALLY IS (AND ISN'T) We are not looking for an "ivory tower" architect or a hands-off engineering manager. We need a highly skilled, pragmatic engineer who is still deeply in love with writing code and shipping systems. We are setting a clear goal—own a learning system that gives our agents opinionated, evidence-backed workflows—and trusting you to own the how. That explicitly includes challenging our assumptions and weighing alternative approaches to get there: we want the strongest architecture, not a predetermined one. The role is split into two primary responsibilities: - 80% Engineering & Building: You will spend the vast majority of your time architecting, writing, and shipping production-ready code. You will inherit a seeded prototype of our knowledge layer and harden it into a robust, scalable, and resilient production platform. - 20% Technical Leadership & Shielding: You will partner closely with the Senior Leadership Team to ruthlessly prioritize the technical roadmap. You will guide other engineers on architectural standards and act as a protective buffer—keeping them safe from the daily "noise" of a fast-growing startup so they can focus on deep, uninterrupted builder mode. WHAT YOU'LL BUILD You will design, own, and scale the architecture behind a continuously learning platform. Specific areas of focus include: - Event collection architecture & customer interaction pipelines to capture rich interaction logs cleanly. - Outcome measurement frameworks to tie AI suggestions to actual business outcomes (sales, retention, clicks). - Recommendation & feedback loops that let the AI automatically improve its behavioral models based on real evidence. - Knowledge graphs, vector databases, and memory/retrieval systems that serve as our persistent cross-product intelligence. - Experimentation infrastructure & feature stores to run secure experiments and manage features efficiently. - Model independence & deployment strategy to architect the system so we can run our own or open-source models on our own infrastructure where it makes sense—a deliberate lever to stay independent of any single LLM provider on both capability and cost. - Evaluation frameworks to continuously benchmark and validate prompt and model improvements. THE HARD PROBLEMS WE'RE BETTING ON We will be honest with you: this is a high-risk, high-reward bet, and part of what makes it worth doing is that the core problems are genuinely unsolved. The central question you will help us answer is deceptively simple—can we reliably tell whether something our AI did led to a better business result? Getting there means confronting a few hard problems head-on, and we would rather debate them openly with you than pretend they do not exist: - Causation, not just correlation: knowing what actually worked, and separating the AI's contribution from everything else happening in a business. - Capturing the outcome: much of the success that matters—a meeting booked, a deal won, a customer retained—happens outside our systems and often isn't tracked today. Instrumenting reliable outcome signals is a first-class part of this role. - Transfer across businesses: what works for a life-insurance broker probably isn't what works for a roofer. We need to learn which know-how generalises and which is context-specific, rather than assuming one business can simply teach another. - Enough signal to learn from: building the data foundations and instrumentation so the loop has enough high-quality evidence to improve—expect meaningful groundwork here before the compounding effects kick in. None of these are reasons not to build it. They are the reasons this role exists, and why we are hiring someone with real learning and recommendation-systems experience to own the architecture that answers them WHAT SUCCESS LOOKS LIKE Within 12 months, you will have shifted us from manual prompt tuning to an automated, compounding loop of intelligence: - Structured Learning: Every customer interaction automatically translates into structured, usable learning data. - Measurable Performance: Every single AI recommendation can be tracked and measured against real-world business outcomes. - Compounding Defensibility: Every experiment run by one customer improves future recommendations for all other customers safely and securely. - Opinionated AI: Our AI agents become increasingly opinionated, moving beyond basic prompt rules to act on real-world evidence of what works. - Autonomous Improvement: The platform improves continuously over time without requiring manual developer intervention. WHO WE'RE LOOKING FOR We value mindset over specific job titles. You are an exceptional systems thinker who thinks in feedback loops rather than simple product features. You naturally ask yourself: "How does this system get smarter every day?" Above all, you have built learning or recommendation systems that measurably improved from real-world feedback—that experience is the anchor for this role. Ideal candidates bring experience in: - Distributed systems, event-driven architecture, and large-scale event processing. - Data engineering, stream processing, feature stores, and robust data modeling. - Graph databases (knowledge graphs) and vector databases for retrieval and memory systems. - Python, TypeScript, SQL, and modern cloud infrastructure (AWS/GCP/Azure). - Recommendation engines, personalization platforms, or reinforcement learning pipelines. - Designing LLM application architectures and robust AI evaluation frameworks (prior GenAI experience is highly beneficial but not strictly mandatory). OUR CULTURE We are a lean, ambitious team that values builders who think deeply but move fast. Our core engineering values are: - Curiosity over Certainty: We ask "how does the system get smarter?" rather than assuming we have all the answers. - First-Principles Thinking: We break complex systems down to their fundamental truths to build elegant, novel solutions. - Shipping over Perfection: We believe working software in production teaches us infinitely more than beautiful designs on a whiteboard. - Long-term Compounding over Short-term Optimization: We design systems that build value over years, not just weeks. - Strong Opinions, Loosely Held: We debate fiercely based on data, but commit fully once a direction is set. - Intellectual Honesty & Ownership: We own our mistakes, speak truth to data, and take absolute responsibility for our outcomes. THE OPPORTUNITY If successful, your work won't just improve an AI product; you will help build a completely new category of business software—one that naturally gains a massive competitive advantage with every company it serves. The learning platform you create will become the foundation of one of the world's most valuable, proprietary datasets on how small businesses successfully operate, grow, and scale. This is a rare opportunity to join as a founding engineer, write a massive amount of core infrastructure, and shape the strategic technical direction of a company on a high-growth trajectory. It is also a chance to own the know-how at the heart of our defensibility—and the option to run our own models—rather than renting that intelligence from someone else. THE INTERVIEW PROCESS We respect your time. Rather than standard algorithm puzzles, we focus on practical systems thinking and collaborative design: 1. Initial Conversation: A casual talk with Kim and Pravin to align on vision, culture, and goals. 2. Architecture Design Exercise: A collaborative, whiteboard-style session focused on designing a real-world learning loop—including how you would prove causation on a first problem like onboarding and retention, and where you would weigh alternative approaches. 3. Technical Workshop: Hands-on programming and collaboration with our core engineering team. 4. Leadership Interview & Strategy Discussion: A deep-dive discussion on product strategy, team dynamic, and long-term vision—including a genuine debate on the trade-offs of this bet: model independence, defensibility, and the alternative paths to owning our know-how.