20 Nov
|
Stealth Startup
|
India
20 Nov
Stealth Startup
India
AI Engineer – Agentic AI & Workflow Engineering
Experience:
3–4 years of total relevant experience
Location:
Dubai, UAE Department: Data Science & AI Innovation Employment Type: Full time We are looking for a hands-on AI Engineer to develop and deploy enterprise AI applications, with a strong focus on Google Agent Development Kit (ADK), AI agents, and automated business workflows. The role combines LLM application development with strong Python engineering, practical system design, and foundational DevOps skills. You will translate business requirements into reliable AI services, integrating models, tools, enterprise data, and human-review steps. You should be comfortable taking a feature from design and implementation through testing, cloud deployment, and production troubleshooting, working closely with senior engineers, data scientists, and business teams.
Key Responsibilities
• Develop AI Agents and Workflows Build single-agent and multi-agent applications using Google ADK. Implement sequential, parallel, conditional, and iterative workflows, combining LLM-driven tasks with deterministic Python logic. Manage agent delegation, session state, and context across workflow steps.
• Engineer LLM and RAG Applications Develop retrieval-augmented generation (RAG), document-processing, and structured-extraction pipelines. Implement prompt and context management, embeddings, retrieval, reranking, and schema-validated outputs. Improve response relevance, grounding, latency, and cost through systematic testing.
• Design Reliable Backend Services Build modular APIs and services using Python and FastAPI. Contribute to system design covering service boundaries, API contracts, data models, asynchronous processing, background jobs, caching, and concurrency. Implement timeouts, bounded retries, and safeguards against duplicate processing.
• Integrate Tools and Enterprise Systems Develop reusable agent tools that connect to databases, document repositories, and business APIs. Use function calling and, where appropriate, Model Context Protocol (MCP) integrations, with clear input schemas, authentication, access controls, and error handling.
• Implement Human Review and Safety Controls Build approval checkpoints and escalation paths for sensitive actions or outputs requiring business validation. Support workflows that pause for human input and resume with the appropriate state. Apply input validation, output checks, and least-privilege tool access.
• Evaluate and Monitor AI Workflows Create representative evaluation datasets and regression tests for output quality, tool selection, and workflow completion. Use logs, traces, and metrics to investigate failures and monitor latency, token usage, and cost. Validate changes to prompts, models, and workflows before release.
• Support Cloud Deployment and DevOps Containerize applications using Docker and deploy AI services to cloud environments, preferably GCP Cloud Run. Maintain basic CI/CD pipelines, environment configuration, secrets, health checks, and rollback procedures in collaboration with platform teams. Google’s ADK deployment guidance supports container-based deployment to Cloud Run. Required Qualifications and Skills Education and Experience
• Bachelor’s degree in Computer Science, Engineering, Data Science, or a related discipline, or equivalent practical experience.
• 3–4 years of total experience focussed in AI engineering, backend development, or software engineering, including hands-on delivery of LLM or agent-based applications.
• Demonstrable experience deploying and supporting applications beyond notebooks or proof-of-concept demonstrations. Core Technical Skills
• Python and backend engineering Strong Python programming, asynchronous execution, REST APIs, FastAPI, Pydantic, exception handling, and maintainable code organization.
• Google ADK and agent development Hands-on experience building agents, custom tools,
and multi-step workflows; working knowledge of sessions, state, context management, callbacks, and agent execution.
• LLM application engineering Practical experience with model APIs, prompt and context engineering, function calling, structured outputs, embeddings, and RAG.
• System design Ability to explain an end-to-end application design and justify choices around APIs, storage, queues, caching, scalability, security, and failure recovery.
• Data and integrations Working knowledge of SQL, REST integrations, document-processing pipelines, and at least one vector database or vector-search service.
• DevOps fundamentals Docker, Git, Linux basics, CI/CD, environment configuration, secrets management, and cloud deployment troubleshooting.
• Testing and debugging Unit and integration testing, API debugging, log analysis, and evaluation of AI outputs and workflow behavior. Preferred Skills
• Experience with ADK or LangGraph or comparable orchestration frameworks, alongside familiarity with graph-based workflows and persistent execution state.
• Exposure to MCP for tool integration and Agent2Agent (A2A) for communication between independently deployed agents.
• Familiarity with Vertex AI, Cloud Run, Cloud Storage, BigQuery, or MongoDB Atlas, and streaming interfaces such as server-sent events or WebSockets.
• Exposure to infrastructure as code, Kubernetes, multimodal document processing, or AI security tooling. Advanced platform administration and model fine-tuning are not core requirements. Working Style We value engineers who break ambiguous requirements into manageable tasks, communicate design trade-offs clearly, and take ownership of implementation quality. You should be comfortable collaborating across teams, participating in code reviews, documenting your work, and learning new AI capabilities without adding unnecessary architectural complexity. The ideal candidate can design, build, deploy, and troubleshoot a practical AI workflow—and explain how its quality, reliability, security, and operating cost will be measured.
📌 Artificial Intelligence Engineer (India)
🏢 Stealth Startup
📍 India