CareerPlanGet AI match score →

Senior Developer Ai Powered Creative Platform

💼 Full-time🗓 2026-07-24

Core

Designing the core architecture for an AI-powered video and image creation platform where users generate UGC-style ads via a schema-driven mode system and AI copilot.

Role type

Senior IC Full-Stack Developer (AI Platform Architecture)

Builds

A scalable, schema-driven creative studio platform integrating multiple third-party AI models for real-time media generation.

Domain

Generative AI, Creative Technology, Video/Image Generation

Deliverable

production ML models | product features

Required skills

System design, Schema-driven architecture, LLM tool-calling, Agentic workflows, Real-time event streaming, Multi-API integration

Preferred skills

Creative tool development, Video/Image model API integration, Plugin system design

Technologies

SSE, WebSockets, LLM APIs (Seedance, Kling, Veo, DALL·E, Midjourney)

Responsibilities

Architect schema-driven mode systems, build AI copilot orchestration layers, integrate multi-model generation APIs, implement real-time progressive updates, unify manual UI and copilot code paths.

Seniority

Senior, hands-on IC

Rewrite
## About the role Speel is building an AI-powered video and image creation platform. Users generate UGC-style video ads — unboxing sequences, talking-head reviews, product demos — by working across multiple AI models in a mode-based workspace. Think of it as a creative studio where AI does the heavy lifting, but the user stays in control. We're looking for a senior developer to own the core platform architecture and ship it. ## What you'll build - A schema-driven mode system where adding a new AI model or creative mode is a config change, not a code change. The schema drives the UI, the AI tooling, the gallery, and the input system — one source of truth. - An AI copilot layer that orchestrates the workspace via tool calls — switching modes, configuring settings, generating media — with a confidence model that decides when to auto-execute vs. ask the user. - Multi-model orchestration across image and video generation models (Seedance, Kling, Veo, DALL·E, Midjourney, and more). Each has different APIs, capabilities, and pricing. Models come and go fast — the integration layer must make adding one trivial. - Real-time progressive updates — when the AI generates 6 images, they appear one by one as they complete, not all at once. Mode switches, status messages, and generation progress stream live. - A dual interaction model — the entire platform works fully without the AI copilot. Manual UI and copilot tool calls share the same code paths and state mutations. No divergence. - Prompt workflow execution — users write multi-step instructions ("generate product photos, then make an unboxing video using those as start frames") and the system decomposes, sequences, and executes across modes with dependency handling. ## What we're looking for - Strong system design instincts. You think in schemas, interfaces, and clean abstractions — not in if-else chains. - Deep experience with LLM tool-calling, agentic workflows, and prompt engineering. You've built systems where AI calls tools, not just answers questions. - Comfort with real-time architectures — SSE, WebSockets, event-driven UI updates. - Experience integrating multiple third-party AI/ML APIs with different interfaces, rate limits, and failure modes. - You ship fast and make pragmatic trade-offs. We're a startup, not an enterprise. ## Nice to have - Experience building creative tools, media platforms, or AI-assisted workflows. - Familiarity with video/image generation model APIs (Runway, Kling, Stability, Replicate, etc.). - You've designed plugin or schema-driven systems where extensibility was a core requirement. ## Details - Role type: Full-time - Compensation: Competitive salary - Stage: Early — you'll have outsized impact on the product and architecture
Sourced via wellfound · Listed on CareerPlan, which tracks 70,000+ jobs from 20+ sources.
Apply on Wellfound ↗