Senior Frontend Software Engineer, AI UX

Pebl•Palo Alto, CA

21h•Hybrid

About The Position

Pebl's mission is to simplify global employment, enabling companies to hire, manage, and retain talent anywhere in the world. We are not just adding a chatbot to our product — we’re rethinking what it means for software to work with you, not just for you. Our AI & Knowledge team is on the front lines of that shift, building Alfie, our conversational AI assistant, and we’re looking for a Senior Full-Stack Software Engineer who wants to help define what AI-native user experience looks like — not follow a playbook that already exists, because it doesn’t yet. We’re looking for someone who is strongest in frontend/product engineering and excited about building conversational AI interfaces that real users interact with directly. You’ll help shape the experience around streaming responses, wrong-answer recovery, fallback states, user trust, and human-in-the-loop moments. This is not primarily a backend AI infrastructure, RAG platform, internal automation, or AI-assisted developer tooling role; the core need is someone who can help drive the user-facing AI experience.

Requirements

8+ years of professional software engineering experience, full-stack, with demonstrated strength and a clear preference for frontend/UX work.
1+ years of hands-on experience building LLM-powered or agentic products — not just calling an API, but designing how an agent reasons, uses tools, and hands control back to the user.
Experience building interfaces for AI-powered or agentic products — conversational UIs, copilots, assistants — including exposure to generative/dynamic UI techniques where the interface adapts to model behavior.
Solid understanding of LLM design patterns, prompt engineering, and observability/instrumentation for agentic systems.
Strong command of modern frontend (React/TypeScript) paired with a solid understanding of how frontends integrate with backend services, APIs, and real-time/streaming data.
A track record of helping build products beyond standard CRUD apps and dashboards — something with real craft, polish, or a distinctive user experience.
Comfort working across the full stack when needed (backend, infra) — this isn’t a pure frontend role.
A lean mindset — focus on delivering customer value by minimizing waste, optimizing workflows, fostering continuous improvement, and prioritizing quick iteration with quick feedback cycles.
Excitement for working in a space that’s evolving rapidly, where the best patterns for AI-driven UX are still being discovered — and a desire to help discover them.

Responsibilities

Design and build full-stack features for our conversational AI assistant, with particular depth and ownership on the frontend/UX side.
Apply hands-on experience with LLM-based and agentic systems — prompt design, tool/function calling, multi-step agent reasoning, evaluation — to shape both how the assistant behaves and how that behavior surfaces in the UI.
Bring product taste and a strong point of view on what makes an AI-driven experience great — push for a better UX rather than shipping “good enough.”
Push the boundaries of what conversational and generative AI interfaces can feel like — prototype and propose new interaction patterns as the underlying models and agent capabilities evolve.
Help shape the future of how people interact with software at Pebl — you’ll be working at the edge of what’s possible in conversational and agentic UX, not maintaining what’s already been figured out.
Build and maintain both frontend and backend components (TypeScript/React on the frontend; comfortable enough on the backend to design clean contracts and APIs that support agentic workflows).
Lead by example in clean, maintainable code across the stack; mentor junior team members and foster a culture of constant learning and improvement.
Contribute to technical discussions, design reviews, and code reviews to maintain high engineering standards.
Troubleshoot issues end-to-end (frontend and backend) in production, including AI-specific failure modes (e.g., bad model output, latency, dropped tool calls).