Job Summary

The Senior Data Architect (R5) is a hands-on leader who designs and personally builds AI and agentic solutions that operate at enterprise scale across Lilly's Clinical and Non-Clinical data domain - multi-agent systems, LLM-orchestrated pipelines, and retrieval/reasoning architectures built on governed, semantic data foundations. This is a builder role: the architect writes code, stands up agent frameworks, and ships production AI systems personally, not just specifications. This role is split 70% hands-on technical execution including coding and 30% strategy, and shapes the future technology landscape.

A defining mandate of this role is Right Model, Right Task - routing every agentic and LLM workload to the model best suited to it on cost, latency, and accuracy grounds, integrated directly with Lilly's internal data platform so routing and reasoning are grounded in enterprise-native lineage and knowledge graphs rather than bespoke, disconnected metadata. Agentic and AI-assisted ways of working are the expected default across every design, analysis, and documentation activity - and this leader is the reference point for how the broader India team scales AI-native architecture.

Path/Level: R5 (Senior Data Architect)

Key Responsibilities

AI & Agentic Solution Architecture (Hands-On)

- Architect and personally build multi-agent and LLM-orchestrated solutions - planning/tool-calling agents, retrieval-augmented generation (RAG), and agent-to-agent workflows - for Clinical and Non-Clinical use cases.
- Design for scale from the start: agent orchestration, state management, concurrency, cost/latency budgets, and failure/retry handling across high-volume production workloads.
- Implement evaluation, guardrails, observability, and human-in-the-loop patterns so agentic systems are safe, auditable, and production-ready in a regulated setting.

Right Model, Right Task - Platform-Integrated Routing, Lineage & Knowledge Graphs





- Design and implement a right-model-right-task routing layer that selects the optimal model - by size, provider, and fine-tuned vs. general-purpose - for each agentic task based on complexity, cost, latency, and accuracy requirements, rather than defaulting to one model for every job.
- Integrate directly with Lilly's internal data platform (Data Hub catalog, lineage, and metadata services) rather than building parallel metadata stores, so routing and agent reasoning are grounded in the enterprises single source of truth.
- Build and maintain lineage-aware enterprise knowledge graphs sourced from the internal data platform, capturing data provenance, sensitivity, and domain context that ground both agentic reasoning and model-selection decisions.
- Use lineage and knowledge-graph context to enforce routing policy - for example, regulatory-sensitive Clinical data routes only to approved/validated models, while low-sensitivity exploratory tasks use lighter-weight, lower-cost models.
- Continuously benchmark and re-tune the model portfolio and routing rules as new models become available, avoiding both over-provisioning expensive frontier models and under-serving tasks that genuinely need them.

AI-Native Data Engineering & Semantic Modeling

- Use AI-assisted and agentic tooling by default across the data lifecycle - pipeline generation, schema/ontology alignment, data-quality scoring, and documentation - not as an occasional accelerator.
- Design conceptual, logical, physical, and semantic models - ontologies, taxonomies, knowledge graphs - that ground both self-service analytics and agentic reasoning, built on and reconciled with the platforms native lineage.




- Build analytics-ready dimensional models and semantic/vector layers that power self-service BI and AI-driven consumption for scientists, statisticians, and business analysts.

Automation & Team Reusability at Scale

- Build reusable agentic accelerators - agent templates, orchestration patterns, evaluation harnesses, model-routing configs - that the broader India team adopts directly to scale AI-native delivery.
- Automate lineage capture, schema-drift detection, and data/agent quality checks so the team spends more time on judgment calls, less on manual upkeep.
- Stand up and maintain the AI/agentic standards platform - the enterprise reference for how agentic solutions, model routing, and knowledge graphs are designed, governed, and delivered across Clinical and Non-Clinical squads.

Governance & Responsible AI

- Implement role-based and attribute-based access control, encryption, and prompt/data-leakage safeguards for agentic systems operating on regulated data.
- Embed data quality, lineage, model/agent risk, and compliance requirements directly into every agentic design and routing decision.

Collaboration & Stakeholder Engagement

- Partner with business SMEs, solution architects, platform/data-catalog teams, and engineering teams to translate Clinical and Non-Clinical needs into scaled agentic solutions.
- Communicate technical trade-offs and AI/agentic architecture recommendations clearly to technical and non-technical stakeholders, and shape the India AI/Agentic roadmap.

Technical Skills & Qualifications

Required - AI & Agentic Solutions at Scale (Must-Have, Hands-On)

- 5+ years hands-on building and shipping AI/agentic systems in production: multi-agent orchestration

Disclaimer: This job posting has been aggregated from external source. Role details, content, and availability are subject to change. Applicants are advised to confirm the latest information directly on the company website before applying.

📌 Senior Data Architect - AI & Agentic Solutions (Bengaluru)
🏢 Eli Lilly
📍 Bengaluru