Solutions Architect - Backend (Agentic AI & Python)

Responsibilities:

• Agentic AI and LLM Orchestration: Design, build, and take to production agentic AI systems with multi-step, tool-using workflows in which LLMs reason, plan, and trigger external function calls to complete complex tasks reliably. Architect and implement robust orchestration and tool-calling loops using frameworks such as LangGraph, AutoGen, or CrewAI, or well-structured custom implementations. Design conversational and workflow state management, memory, context handling, and multi-turn continuity for reliable long-running agent interactions.
• Implement retrieval-augmented generation (RAG) patterns: document ingestion, chunking, embeddings, vector stores, and grounded generation with citations to improve accuracy and traceability.
• Apply advanced prompt engineering, guardrails, evaluation, and fallback strategies to maximise reliability and minimise hallucination in production agent behaviour.
• Integrate model APIs (such as OpenAI and Anthropic) into backend services, focusing on the agentic routing and orchestration layer rather than training foundational models.
• Design agent-to-agent and multi-agent orchestration patterns for complex, decomposed workflows.
• Backend Architecture and API Development: Architect, design, and develop robust, scalable backend services and RESTful/event-driven APIs that power AI-driven and conventional workloads at scale. EngiNeo Solutions
• Lead the end-to-end design of backend systems from domain modelling and service.
• Build and maintain services using modern Python frameworks, with deep expertise in FastAPI, applying clean architecture and well-established design patterns.
• Own service performance tuning, profiling, query optimisation, caching, and resource management to ensure efficient handling of large-scale traffic.
• Design and implement asynchronous and concurrent processing patterns using async/await, background workers, task queues, and message-driven workflows.
• Build real-time streaming interfaces using Server-Sent Events (SSE) or WebSockets to stream LLM responses and live updates to clients.
• Implement reliable messaging and event-processing patterns using tools such as Kafka, RabbitMQ, or Celery where applicable.
• Define reliability standards, retry logic, timeout handling, graceful degradation, and idempotent processing across all asynchronous and AI workloads.
• Third-Party Integration and Orchestration: Design and build integrations with external APIs and platforms, managing complex multi-step orchestration flows and service-to-service state machines.
• Handle asynchronous third-party failure states gracefully alongside the agent's reasoning logic, with robust timeout, fallback, and reconciliation strategies.
• Build integration and middleware layers that abstract external dependencies and keep the core system decoupled and testable.