solution architect for in-vehicle LLM backend

The client is advancing an in-vehicle voice assistant into an AI-powered companion. Its cloud AI orchestration service receives requests from vehicles, routes them to LLMs, tools, and agents, and returns answers or actions to the car. DXC Luxoft serves as the end-to-end delivery partner, working with the client’s engineers on the Azure platform.

• Own and evolve the backend service architecture, including routing, business logic, orchestration, service decomposition, and structural design for long-term changeability
• Define and govern the interface contracts between vehicles, LLMs, tools, and agents, including versioning and evolution across vehicle generations
• Architect vehicle-facing integrations, including REST and gRPC/protobuf APIs, Viwi, AIDL, OAuth2, mTLS, client IdP integration, and internal host and telemetry interfaces
• Design streaming architecture for incremental chat completion, real-time ASR transcription, and TTS playback during synthesis, including buffering, connection management, error handling, cancellation, and interruption
• Own the LLM integration layer, including prompt normalization and processing, response formatting for vehicle displays, context preparation, dialogue management, and deterministic vehicle answers through system prompts and guardrails
• Design agent orchestration for routing requests to the LLM, internal tools, or external agents, including multi-intent handling and the LangGraph routing model
• Architect tool and agent integrations for navigation, media and entertainment search, knowledge queries, calendar and mail, POI and places services, and vehicle data services
• Define RAG and persistence architecture, including embedding strategy and lifecycle, vector store design, schema design, tuning, migrations, and data-store responsibilities
• Develop scaling and load concepts for series operation, plus security, data protection, and automotive compliance concepts, including data minimization in telemetry and traces
• Own end-to-end observability architecture using distributed tracing, prompt tracing and evaluation, metrics, dashboards, alert thresholds, and routing
• Ensure the architecture meets operational targets, including 99.5% SLO availability, incident resolution within 24 hours, business-hours second- and third-level support, and structured version, release, and deployment management
• Guarantee backward compatibility for existing vehicle generations, including vehicles from model year 2021 onward
• Set and enforce engineering standards with development teams, including API and interface reviews, Python code quality gates, CI/CD pipeline design, and Terraform/Terragrunt IaC module structure
• Act as the technical counterpart to client architects and vehicle, backend, and UX teams; maintain architecture documentation, work with the client’s requirements management tools, and provide technical direction to the distributed development and AI Ops team in Scrum

• 8+ Years of backend or distributed systems engineering experience, including 3+ years as an architect or technical lead with end-to-end ownership of production system design
• Expert Python skills, including FastAPI and async, with hands-on experience in streaming architectures such as SSE, WebSocket, or gRPC streaming, back pressure, and cancellation
• Proven experience architecting production LLM systems, including tool- and agent-based orchestration, prompt strategies, structured outputs, constraint handling, and guardrails
• Production experience with Azure OpenAI Service or OpenAI API, including model version migrations and compatibility challenges
• Strong API and interface design skills, including REST, gRPC/protobuf, versioning, backward compatibility, OAuth2, and mTLS
• Deep PostgreSQL experience, including schema design, tuning, and Alembic migrations; experience with embeddings and vector search for RAG
• Kubernetes and container orchestration experience, including deployment, scaling, secrets and configuration, networking, and load balancing; Azure architecture experience with AKS or Container Apps, Managed Identity, Key Vault, and Monitor
• Terraform Infrastructure as Code experience, including reusable modules, environment separation, remote state and locking, CI/CD integration, and lifecycle management
• Demonstrated responsibility for operating production services, including monitoring, alerting, OpenTelemetry tracing, incident and problem management, patch and release management, and latency and cost optimization against defined targets
• Будет плюсом: automotive or in-vehicle software, MIB3/E³/SDV platforms, Android Automotive and AIDL, Viwi, vehicle telemetry and configuration services, SOA/microservice/zonal E/E architectures, LLM evaluation and prompt regression, LangFuse, DeepEval, Ragas, PromptFlow, faithfulness and hallucination metrics, latency P95, WER/CER, streaming ASR/TTS integration, in-cabin speech conditions, HumeAI and empathic dialogue design, MongoDB and CosmosDB, semantic caching and token-cost optimization at fleet scale, automotive quality/security/AI governance frameworks, and taking over and stabilizing a running system from a previous supplier

• Условий в вакансии нет