About the role
Responsibilities:
- Agentic code generation system with multi-step tool-calling from design inputs.
- Code editing pipelines that apply intelligent multi-file edits from natural language.
- Design-to-code engine converting Figma designs to production React components.
- Code quality infrastructure with automated benchmarking and production feedback loops.
Requirements:
- 5+ years software engineering; 2+ years hands-on experience with LLM APIs, prompt engineering, and production deployments for code-related tasks (code generation, editing, transformation, or developer tools).
- Demonstrated experience building code generation, code editing, or code transformation features.
- Strong proficiency in TypeScript/Node.js and Python.
- Experience with agentic AI systems: tool calling, multi-step reasoning, and error recovery in code generation contexts.
- Understanding of model evaluation and benchmarking for code quality.
- Hands-on builder who can architect and code, not just provide direction.
Nice-to-Have Skills:
- Fine-tuning LLMs for code tasks (LoRA, adapters, distillation).
- Code analysis tools: tree-sitter, LSP, AST manipulation.
- Real-time streaming architectures (WebSocket) and queue systems (BullMQ, Redis).
- Model optimization: ONNX, quantization.
- Prior experience in the design-to-code or developer tools domain.