AI Developer - Mid Level
Job Description
Job DescriptionPosition Summary: Title: Developer Premium IIIDuration: Long TermLocation: Washington, DC Hybrid Onsite: 4 days per week from Day 1, with a full transition to 100% onsite anticipated soon Job Requirement:
Should have at least 6+ years of Development experience with Python, React and other Development tools.
AI-first engineering mindset: Able to decide when to use traditional code versus large language models or AI agents. Designs for probabilistic AI behavior, fallback approaches, human review, and cost, latency, and reliability trade-offs. Generative AI development: Hands-on experience integrating large language models into production applications, prompt engineering, structured outputs, retrieval-augmented generation, embeddings/vector search, context management, and evaluation for hallucination and response quality. Rapid AI solution development: Comfortable with modern AI development platforms, APIs, and tooling to rapidly prototype, iterate, and deploy working AI capabilities into real systems. Agentic AI: Experience designing agents with goals and constraints, orchestration patterns, tool/API calling, state and memory management, guardrails, failure recovery, and behavioral monitoring. Full-stack engineering: Backend experience with REST/GraphQL APIs, secure services, data models, asynchronous processing, and enterprise integrations; frontend experience with modern frameworks such as React and AI-specific user experiences such as streaming responses, explainability, review/edit/retry/approve interactions. Cloud-native development: Containers and managed services, continuous integration and deployment for AI components, secure secrets/key management, performance optimization, and cost-aware AI design. Enterprise AI reliability and security: Sensitive-data protection, data leakage risks, bias/safety/misuse considerations, auditability, traceability, and graceful degradation when AI fails. System thinking: Break ambiguous problems into components, build end-to-end systems rather than demos, balance rapid iteration with maintainability, and think in architectures.
