17 Nov
|
Yamaha Motor Solutions India
|
Faridabad
17 Nov
Yamaha Motor Solutions India
Faridabad
Enterprise GenAI Solutions | LLM Architecture | Agentic AI | Production EngineeringExperience: 3–6 years overall, including a minimum of 3 years of hands-on Generative AI solution engineering experience
Role Level: Senior AI Engineer / Lead GenAI Engineer (Individual Contributor)
Primary Focus: Architect, engineer, deploy, and scale enterprise-grade Generative AI applications
Mandatory: Proven production deployment experience. Proofs of Concept (PoCs), prototypes, hackathons, and experimental trials will not qualify toward the mandatory 3 years of production GenAI experience.
A. GenAI Solution Architecture & Engineering
- Design end-to-end enterprise GenAI application architectures covering UI, API gateways, orchestration, model inference, retrieval, enterprise data integration, observability, security, and infrastructure.
- Build production-grade LLM-powered applications for enterprise knowledge management, intelligent document processing, engineering assistants, workflow automation, and decision support.
- Design architectures supporting multi-tenancy, high availability, fault tolerance, scalability, extensibility, and disaster recovery.
- Implement structured outputs, function/tool calling, context engineering, memory management, and intelligent model routing.
- Evaluate architectural trade-offs among hosted proprietary models, open-weight models, self-hosted inference, and hybrid deployments.
- Design model-agnostic integration layers to minimize vendor lock-in.
- Apply asynchronous processing, caching, batching, rate limiting, and fallback strategies to optimize production applications.
B. Advanced RAG & Enterprise Knowledge Systems
- Architect and implement advanced RAG systems with production-ready ingestion, indexing, retrieval, generation, and evaluation.
- Develop document ingestion and processing pipelines for PDFs, Office documents, HTML, structured data, and scanned content.
- Implement semantic chunking, metadata-aware indexing, embedding strategies, hybrid search, and reranking.
- Design query rewriting, multi-query retrieval, hierarchical retrieval, and advanced retrieval patterns where appropriate.
- Integrate vector databases and enterprise search platforms.
- Ensure document-level and user-level access controls are preserved during retrieval.
- Establish mechanisms for source attribution, grounded answers, hallucination reduction, and retrieval quality monitoring.
- Manage document versioning, re-indexing, incremental ingestion, and data freshness.
C. Agentic AI & Workflow Orchestration
- Design and implement multi-step, tool-using AI agents with explicit workflow orchestration and state management.
- Build controlled agentic workflows using LangGraph, equivalent orchestration frameworks, or custom implementations.
- Integrate AI agents with enterprise APIs, databases, business applications, and automation platforms.
- Implement guardrails, human-in-the-loop approval gates, tool authorization, and escalation mechanisms.
- Apply bounded autonomy, execution limits, timeout handling, idempotency, retries, and compensating transactions.
- Maintain auditable decision traces, tool execution records, and explanations for consequential agent actions.
- Evaluate the suitability of deterministic workflows versus agent-based architectures based on business risks and requirements.
D. Production Engineering, LLMOps & Scalability
- Develop scalable backend services using Python, FastAPI, REST APIs, and asynchronous programming.
- Package and deploy applications using Docker, CI/CD pipelines, and cloud-native infrastructure.
- Establish production-grade LLMOps practices covering model configuration, prompt versioning, evaluation gates, deployment, monitoring, and rollback.
- Implement observability across application traces, model calls, retrieval operations, agent execution, and infrastructure.
- Optimize p50/p95/p99 latency, throughput, token consumption, inference cost, and infrastructure utilization.
- Implement load testing, performance benchmarking, failure recovery,
and production incident management.
- Support cloud, hybrid, and on-premises deployment environments.
- Define service-level objectives (SLOs), operational runbooks, and production support processes.
E. AI Quality, Evaluation & Responsible AI
- Develop automated evaluation pipelines for retrieval performance, answer faithfulness, relevance, completeness, and task success.
- Build representative golden datasets and regression test suites.
- Implement human evaluation and LLM-as-judge methodologies with appropriate calibration and validation.
- Evaluate agents for tool-selection correctness, workflow completion, policy compliance, and unsafe actions.
- Implement controls against prompt injection, sensitive data leakage, unauthorized tool execution, and insecure retrieval.
- Apply enterprise IAM, RBAC/ABAC, encryption, secrets management, audit logging, and data retention requirements.
- Work within applicable AI governance and security standards, including relevant practices from ISO/IEC 42001 and OWASP guidance for LLM applications.
F. Forward Deployed Engineering (FDE) & Stakeholder EngagementA critical requirement of this role is the ability to operate as a Forward Deployed Engineer, combining technical depth with direct business engagement and accountable delivery.
- Work directly with customers and business stakeholders to understand operational challenges and translate them into viable AI solution architectures.
- Lead technical discovery, requirements engineering, feasibility assessment, solution design, implementation, and production deployment.
- Independently investigate complex enterprise environments, legacy applications, data constraints, and integration challenges.
- Develop reusable architectural patterns, accelerators, deployment templates, and AI engineering assets.
- Conduct architecture reviews and explain design decisions to engineering teams, leadership, and non-technical stakeholders.
- Drive adoption, knowledge transfer, user enablement, and operational handover.
- Take ownership of delivered solutions beyond release, including performance, stability, business value, and continuous improvement.
📌 Generative AI Engineer – (FDE) (Faridabad)
🏢 Yamaha Motor Solutions India
📍 Faridabad