About the role
Specright is seeking an Agentic Software Engineer (Full Stack) in Egypt for a 100% remote role. This production-focused position involves delivering dependable, user-facing AI agent products rather than conducting research. The engineer will take complete responsibility for deployed systems, from agent architecture and coding through evaluation, operation and continued improvement.
What you will do
- Design, build and release complete agentic applications for real production users, including conversational interfaces, AI workflow products and internal copilots.
- Use agentic development assistants such as Claude Code, Codex and Cursor each day to speed delivery while preserving engineering quality.
- Architect tool calling, multi-stage orchestration, memory and state handling, retrieval pipelines, permission structures and recovery from failures.
- Create and maintain agent-quality systems that include regression suites, scenario tests, golden traces, and online and offline metrics.
- Add production-grade monitoring through distributed tracing, version and prompt records, tool-call telemetry, and tracking of cost and latency.
- Manage production readiness across QA, rollout planning, guardrails, incident handling and the ongoing refinement of deployed agents.
- Write and review code while operating the systems delivered.
Requirements
- Demonstrated full-stack engineering ability, including end-to-end delivery of production software rather than only prototypes or demonstrations.
- Practical experience creating production LLM agents, with ownership of at least 1-2 agentic products or features through release and operation.
- Deep knowledge of good agentic coding practice, including task decomposition, agent delegation, prompt hygiene, iterative loops, and effective review and steering of agentic output.
- Applied experience with agent evaluations and tracing, including behavioral evaluation systems, trace-driven debugging and regression detection beyond standard model benchmarks.
- Strong engineering foundations spanning testing, debugging, basic security, performance tuning and reliability.
- TypeScript/Node.js and/or Python proficiency, modern web-stack experience and the ability to deliver at startup speed are advantageous.
- Familiarity with agent-tooling environments is desirable, including prompt/version stores, evaluation harnesses, and tracing or telemetry products such as OpenTelemetry, LangSmith, Datadog and Honeycomb.
- Experience with RAG architecture is desirable, covering vector databases, indexing, chunking techniques and retrieval evaluation.
- Experience designing agent-facing tool APIs is desirable, including structured-output schemas, idempotency, rate limits and error-handling methods that improve agent reliability.
Benefits
A 100% remote workplace with an open, relatively flat structure and an emphasis on collaboration, sharing information and exchanging ideas.
How to apply
Apply through Specright's official vacancy.