Full Stack Engineer (React & Python)

About the Role

You build the product experience for M-agentic – Agentic Process Studio – the six-stage wizard that lets an enterprise user turn a process into deployed AI agents. Across the stack, you'll rebuild the front end in React + TypeScript on our design system and build the Python backend services and APIs behind it, wiring the wizard to live Amazon Bedrock (Claude) calls and to the agent runtime for our Health Insurance Claim Adjudication pilot.

Key Responsibilities

• Build the wizard screens in React + TypeScript: process intake, flow/graph visualizations, gap analysis, agent architecture, and the Agent Forge (configure, test-trace, run-demo).
• Build Python backend services: REST APIs, transformation persistence and versioning, authentication, and the Bedrock orchestration endpoints the wizard calls.
• Implement the Forge deploy path – turning wizard configuration into a deployable agent spec (per the Architect's contract).
• Integrate the design system faithfully; own responsive layout, empty/error/loading states and accessibility.
• Replace the prototype's localStorage/mocked data with real, persisted state and live services.
• Write tested code, participate in code review, and collaborate closely with AI Engineers on the API surface between UI and agents.
• Contribute to CI/CD and observability for the services you own.

Required Qualifications

• 2+ years building web applications full-stack.
• Strong React + TypeScript on the front end and Python (FastAPI/Flask/Django or similar) on the back end.
• Solid API design (REST/JSON), relational or document data stores, and authentication patterns.
• Comfortable on AWS (Lambda/containers, S3, IAM basics) and with CI/CD and Git workflows.
• Good eye for UX detail and turning designs into clean, accessible interfaces.

Preferred

• Experience rendering interactive diagrams / flow graphs (SVG, canvas, or a graph library).
• Exposure to LLM APIs (Amazon Bedrock/Claude or comparable) and streaming responses.
• Design-system / component-library experience; prior work on data-heavy enterprise tools.

What Success Looks Like (First 8 Weeks)

• The six wizard stages rebuilt and working against real services – no mocked data.
• The wizard calls Claude for real (intake parsing, gap analysis) and can configure & deploy an agent through the Forge.
• Polished, responsive, accessible UI ready for the pilot demo.