Senior Data Architect

Job Description: Senior Data & AI Architect

Position: Senior Data & AI Architect

Send ur resume to zeusailtd@gmail.com

Role Summary

We're looking for a hands-on architect who can take an organisation's scattered documents, data and code and turn them into a knowledge layer that answers questions accurately, with citations. You'll design how data is cleaned and ingested, how it is retrieved, and how it is structured into ontologies and knowledge graphs. This is a build role: you draw the architecture and you write the code that proves it works.

Key Responsibilities

Data Ingestion & Processing: Design ingestion pipelines for PDFs, Office files, scanned documents, tables, audio, video and source code, keeping source provenance (page, paragraph, table cell, timestamp) on every chunk.

Data Quality & Versioning: Define cleaning, de-duplication, language detection, versioning and incremental update rules so only changed content is re-processed.

Chunking Strategy: Choose chunking approaches per content type and prove them with retrieval metrics, not intuition.

Retrieval & RAG Architecture: Architect hybrid retrieval combining keyword and vector search, rank fusion, reranking, query decomposition and tiered context.

Grounding & Citations: Design grounding checks so the system cites its sources, drops unsupported claims, and asks for clarification when evidence is thin.

Ontology Design: Work with domain experts to define ontologies, including entity types, relationships, naming and versioning rules.

Knowledge Graphs & GraphRAG: Build typed entity and relationship extraction with entity resolution, and decide when answers should come from graph traversal, text passages, or both.

Evaluation & Knowledge Health: Build golden question sets and evaluation harnesses that gate every release; measure retrieval recall, faithfulness, citation accuracy, coverage and freshness.

Governance & Observability: Design access control, audit trails, tracing and cost tracking into the data and retrieval layers.

Agent Integration: Expose the knowledge layer as tools other AI agents can call (search, cite, traverse), including through MCP, and design agent memory that stays separate from curated knowledge.

Model Selection: Benchmark embedding models, rerankers and LLMs across providers and recommend the right mix for quality and cost.

Technical Leadership: Guide a team of data engineers and data scientists, run design and code reviews, and set standards for how retrieval quality is tested.

Client Consulting: Work directly with client architects and business stakeholders, explain trade-offs in plain language, and drive decisions.

Innovation: Stay current with RAG, GraphRAG and knowledge engineering research and bring practical recommendations to the client.

Required Skills & Qualifications

Technical Proficiency:

Strong hands-on experience in Python and SQL.

Hands-on experience building and shipping RAG systems to real users. You should be able to walk us through one you built and how you measured and improved its quality.

Document processing at scale: PDF and Office parsing, OCR, table extraction and layout-aware chunking (e.g., Docling, Unstructured).

Hybrid search: BM25/full-text search, dense embeddings, rank fusion and cross-encoder reranking, using pgvector, OpenSearch, Elasticsearch or a dedicated vector database.

Knowledge graph and ontology design: entity modelling, entity resolution and schema evolution, with Neo4j, RDF/OWL, SPARQL or graph models in relational stores.

Hands-on GraphRAG experience: LLM-based entity and relation extraction, graph and community retrieval, combining graph and passage context (e.g., Microsoft GraphRAG, LightRAG, LlamaIndex).

Practical experience with LLM evaluation (e.g., Ragas, DeepEval or custom harnesses) used as a release gate.

Professional Experience:

A minimum of 10 years in data engineering, data science or search/ML, including 2+ years hands-on with RAG and at least one system taken to production.

2+ years in an architect or technical lead role, owning design decisions and guiding engineers.

Prior client-facing experience in consulting or customer-embedded delivery.

Bachelor's or Master's degree in Computer Science, Data Science, Information Science or a related field.

Additional Skills:

Solid data engineering practices: lineage, idempotent jobs, versioned datasets and incremental processing.

Strong working knowledge of at least one cloud platform (AWS, Azure or Google Cloud) and its AI services.

Experience with LLM APIs (e.g., OpenAI, Anthropic Claude, Amazon Bedrock, Azure OpenAI), including structured outputs.

Hands-on with Docker and CI/CD pipelines.

Awareness of data security and access control in retrieval systems.

Ability to sit with domain experts, learn their vocabulary, and turn it into a working ontology.

Strong verbal and written communication, with the ability to explain retrieval and graph concepts to non-technical stakeholders.

Comfortable handling ambiguity and making sound technical calls independently.

Ability to work collaboratively in an Agile/Scrum environment across time zones