Description
We are looking for a senior solution architect to own the technical vision and lead an engineering team building a modern data, semantic, and AI platform. You will be the technical authority for how data flows, gains semantic structure, trains models, and powers AI agents — leading a multidisciplinary team (data engineers, ML engineers, semantic/ontology engineers, backend) from design to production.
This is a hands-on leadership role. You will set architecture direction, define the contracts between data, semantic, and AI layers, write the patterns the team builds against, and stay close enough to the code to debug a data transform, a GraphQL/policy path, or an agent flow.
Requirements
- Own the end-to-end architecture of the data – semantic – AI stack and the contracts between layers (GraphQL, REST, protobuf, event streams, tool-calling interfaces).
- Lead and mentor the engineering team, reviewing designs and PRs against shared architecture decisions and reference patterns.
- Architect the data platform – batch ingestion, a versioned lakehouse with branch-based safety (staging – quality gates – promotion), distributed transformation and orchestration, dataset versioning, and event-driven pipelines.
- Design the semantic/ontology layer – entity, property, and relationship modeling; semantic query APIs; fine-grained policy enforcement; and dataset/column linkage that gives data meaning.
- Drive the AI and MLOps layer – agent orchestration, AI copilot experiences, model guardrails, LLM inference, and the full experiment-tracking – model-registry – serving lifecycle.
- Guarantee data and semantic quality – quality gates, ontology consistency, embedding/vector strategy, PII handling, and behavior evaluation that gates releases.
- Embed governance by design – data lineage, auditability, multi-tenancy, and policy enforcement as first-class concerns.
- Be the technical voice with stakeholders, translating data-governance and AI-operations requirements into architecture, and running design reviews and demos.
Must-have qualifications
- 10+ years in software, data, or ML engineering, with 5+ years as a solution, data, or platform architect leading teams and owning system-level design.
- Deep distributed-data expertise – lakehouse architecture (e.g., Iceberg, Delta, Hudi), data versioning/branching, distributed compute (e.g., Spark), event-driven pipelines (e.g., Kafka), and data lineage/quality patterns.
- Semantic, ontology, or knowledge-modeling experience – entity-relationship or ontology design, semantic query layers, and GraphQL APIs; familiarity with policy-as-code for fine-grained access control.
- MLOps and AI-agent systems – model registry and serving (e.g., MLflow, KServe), feature stores, LLM inference (e.g., vLLM), agent frameworks (e.g., LangGraph), and guardrails/safety patterns.
- Vector search and retrieval – embeddings, semantic search (e.g., Qdrant, pgvector), and RAG patterns.
- API and contract design – GraphQL, REST, protobuf, and event schemas with strong versioning discipline.
- Hands-on coding in Python across the data/ML stack (e.g., FastAPI, PySpark).
- Strong technical leadership – mentoring engineers, writing architecture decision records, running design and PR reviews, and aligning a team around shared patterns.
- Governance-by-design instinct – lineage, audit, multi-tenancy, and policy treated as first-class, not afterthoughts.
Nice-to-have
- Experience in regulated industries (financial services, healthcare, government) where data governance, lineage, and auditability are requirements.
- Familiarity with tool-calling/agent integration protocols and contracts.
- Background with declarative transformation frameworks (e.g., dbt) and data-quality tooling.
- Working comfort with Kubernetes as a deployment substrate – enough to be effective, even if it is not your primary focus.
- Familiarity with an observability stack (e.g., OpenTelemetry, Prometheus, Grafana) and immutable-audit patterns.
- Prior experience taking a data/AI platform from proof-of-concept to production end-to-end.
Benefits
- Data flows cleanly from source – versioned dataset – semantic layer – intelligent action, with quality and policy gates holding under real workloads.
- The semantic layer becomes the trusted foundation that agents, copilots, and analysts query with confidence.
- ML experiments, models, and agents ship predictably and safely, with lineage and audit intact.
- Architecture boundaries and platform invariants hold as the system scales.