Overview:

The AI Solution Engineer is a delivery focused engineer who designs, builds, and operates production agentic AI applications, along with the platform services those applications run on. Working within the technical direction set by AI Solutions and Platforms leadership, the AI Solution Engineer translates solution architecture into production grade agents, retrieval pipelines, tool and system integrations, evaluation harnesses, and the guardrails and observability that keep those systems reliable once live.

The AI Solution Engineer is a hands on practitioner who codes, integrates, tests, and iterates, owning feature level delivery across one or more solution or platform tracks while building a working understanding of the wider AI platform. The role works daily with modern AI engineering tooling, including agentic frameworks such as LangChain and LangGraph, the Model Context Protocol, and coding agents such as Claude Code, GitHub Copilot and Codex. This is a technical builder first, with a growing architect mindset

Responsibilities:

Responsibilities

- Solution Architecture and Delivery: Architect, build and deploy production agentic AI applications for R&D;, taking them from problem definition and prototype through to production use by scientists and engineers.
- Hands On Engineering: Write production Python and own the difficult parts, including retrieval quality, tool and function design, multi agent orchestration, context management, structured outputs, latency and unit cost. Set the technical bar through design and code review.
- Agentic Framework Selection: Build with LangChain, LangGraph, MCP and agent to agent patterns. Select the appropriate model across Anthropic Claude, Azure OpenAI and Google model families, and recognise when a deterministic or classical machine learning approach is the correct answer.
- Evaluation and Quality: Establish evaluation as a delivery gate, covering golden datasets built with R&D; subject matter experts, automated evaluations in CI, regression and grounded Ness testing, red teaming, and human review in the loop. Take solutions from prototype into production and remain accountable for them once live.
- Platform Ownership: Own the shared AI platform for R&D;, including retrieval and vector infrastructure, agent runtime, prompt and model registry, MCP server and connector layer, guardrails, tracing and observability, cost attribution and rate management. Define reference architectures and reusable patterns so that downstream teams ship in weeks rather than quarters.
- MLOps and LLMOps: Embed CI/CD for models, prompts and agents, infrastructure as code, environment promotion, versioning, monitoring, incident response, model deprecation and migration.
- Enterprise Integration:



Integrate AI solutions with the R&D; and enterprise estate, including PLM, LIMS, specification and document management systems, SAP, the Azure data platform and the enterprise Lakehouse.
- Security and Governance: Own security, privacy and governance for these systems in partnership with Cyber, Legal, Privacy and Responsible AI, covering access control, secrets management, handling of confidential formulation and specification data, and auditability. Contribute to AI governance standards for R&D.;
- People Leadership: Lead, coach and grow a team of AI and platform engineers. Develop technical depth in the team and build AI literacy across the wider R&D; community.
- Stakeholder Partnership: Partner with R&D; scientists, category and platform leaders to convert ambiguous scientific problems into scoped, valuable products, and to decline those that will not work. Manage delivery partners and vendors. Communicate credibly with a food scientist and a cloud architect in the same conversation, and with senior leadership on portfolio, risk and value.

Qualifications:
- Bachelor’s degree in computer science, Engineering, Data Science or a related quantitative field. Masters preferred.
- 10 to 14 years in software, data or machine learning engineering, including at least 2 to 3 years building and running LLM and generative AI solutions in production.
- 2 or more years leading engineers, formally or as a senior technical lead, with evidence of having developed people.
- Strong hands-on Python. Code quality will be assessed during the interview process.
- Deep practical experience with agentic AI, covering tool calling, planning, memory, multi agent and agent to agent patterns, retrieval augmented generation, structured output, guardrails and evaluation.
- Production experience with LangChain and LangGraph, or an equivalent framework such as LlamaIndex, Semantic Kernel, CrewAI, Pydantic AI or custom orchestration, with a transparent view of the tradeoffs.
- Daily working use of AI coding agents such as Claude Code, GitHub Copilot, Codex or Cursor, including agent configuration and custom tooling rather than inline completion alone.
- Solid understanding of LLM fundamentals, including context windows, embeddings, tokenization, fine tuning versus adapters versus prompting, and failure modes such as hallucination and prompt injection.
- Cloud and platform engineering experience. Azure preferred, covering Azure OpenAI or AI Foundry,



AKS and Functions, or the equivalent on AWS or GCP. Docker, Kubernetes, Terraform or Bicep, CI/CD and API design.
- Data foundations, including SQL, Spark or Databricks, vector stores, and real experience working with messy unstructured technical documents.

Overview (Secondary Language):
- Prior delivery of technology for an R&D;, scientific or laboratory function in CPG, food and beverage, pharma, chemicals, flavors and fragrances, agriculture or materials science.
- Working familiarity with scientific and product data, including formulations and recipes, technical specifications, bills of materials, ingredient masters, sensory panel data, nutrition and claims data, and stability datasets.
- Experience with R&D; systems of record such as PLM, LIMS or ELN, and with specification and technical document management.
- Track record of taking agentic systems past pilot into sustained production use, with adoption and reliability data to evidence it.
- Experience building MCP servers and custom tool layers over enterprise APIs and internal systems.
- Retrieval engineering depth beyond baseline RAG, including hybrid and semantic search, reranking, chunking strategy for technical documents, and table and figure extraction.
- Databricks at production scale, including Unity Catalog, Delta and Lakehouse patterns.
- LLM cost engineering, including token accounting, caching, batching, routing across model tiers and chargeback to business units.
- Experience building internal developer platforms or shared services consumed by other engineering teams.
- Exposure to Responsible AI frameworks, AI model risk management, or EU AI Act readiness and comparable emerging regulation.
- Experience handling commercially sensitive or trade secret data, including how intellectual property is protected when third party models sit in the processing path.
- Experience in a global capability center delivering for global stakeholders across time zones, where the hub owned outcomes rather than executed tickets.
- Having hired, levelled and grown AI or platform engineers in the Indian market.
- A postgraduate qualification in Computer Science, Engineering or Data Science, or in a scientific discipline relevant to food, chemistry, materials or nutrition. Candidates who moved from science into engineering are of particular interest.
- Recognized AI engineering credentials, for example Anthropic's Claude Certified Developer Foundations or Claude Certified Architect exams, GitHub Copilot certification, or Microsoft, AWS or Google AI engineering certifications. Equivalent demonstrated experience is weighted equally.
- Open-source contributions to agent or LLM tooling, conference talks, published papers or technical writing.

📌 Deputy Director - AI Solution and Platforms (Hyderabad)
🏢 PepsiCo
📍 Hyderabad