Agentic Systems & Knowledge Graph Engineer for a Strategy AI SaaS Platform

We are looking for a strong AI Engineer to design and build the intelligence layer of our strategy generation enterprise-level platform: agentic systems that reason over a structured knowledge graph to produce auditable, high-quality output. You will own architecture and implementation end to end, from ontology and graph data modeling through multi-agent orchestration on AWS. This is a hands-on engineering role with significant architectural latitude; you will set technical direction, not just execute against it. You should be equally comfortable whiteboarding a data model, debating an orchestration pattern, and shipping the production code that implements both. What You’ll Do • Work with a Solution Architect/Subject Matter Expert to design and implement the knowledge graph layer on Amazon Neptune: schema and ontology modeling, ingestion pipelines, query patterns (Gremlin and/or SPARQL), and performance tuning at scale. • Define and evolve ontologies and taxonomies that give the system a shared, machine-reasonable model of the domain, and keep them coherent as requirements grow. • Architect and build agentic systems: multi-agent orchestration, planning and tool-use loops, retrieval and grounding, evaluation, and guardrails. Reason explicitly about cost, latency, reliability, and output quality. • Build retrieval-augmented (RAG) and graph-augmented workflows that combine semantic retrieval with structured graph traversal. • Produce architecture artifacts (system diagrams, data models, interface contracts, and design docs), and drive them through review to implementation. • Own services on AWS, using managed AI and orchestration primitives (e.g., Bedrock, Step Functions, Lambda, OpenSearch) with sound IAM, observability, and operational practices. • Establish evaluation and quality discipline: deduplication, provenance/evidence tracking, point-in-time snapshots for audit and replay, and regression testing for non-deterministic systems. Required Qualifications • Strong software engineering fundamentals and the ability to ship production code (Python and/or TypeScript/Node.js). • Demonstrated experience designing and operating graph databases (Neptune strongly preferred), including data modeling and one or more graph query languages (Gremlin, SPARQL, openCypher). • Practical experience with ontologies, taxonomies, or semantic data modeling, and the judgment to know when structure helps versus when it adds friction. • Hands-on experience building agentic / LLM-powered systems (orchestration, tool use, retrieval, prompting, and evaluation) that work reliably in production, not just in demos. • Solid AWS experience: able to architect, deploy, secure, and operate cloud services independently. • Ability to create architecture from ambiguity: translate a fuzzy goal into a clear design, communicate it in diagrams and docs, and defend the tradeoffs. Nice to Have • Experience with AWS Bedrock, Amazon Agents and Knowledge Bases, Step Functions, and OpenSearch in a production AI pipeline. • Familiarity with web-scale retrieval and search integration, and fallback scraping strategies. • Background in evaluation frameworks for LLM systems, including cost/quality optimization where query design is the primary lever. • Experience with audit/replay-grade systems where provenance and reproducibility are first-class requirements. Who You Are • You think in systems and care about why a design holds up, not just whether it runs today. • You treat quality, cost, and latency as design inputs rather than afterthoughts. • You write clearly, and your designs are legible to engineers, product, and stakeholders alike. • You move fast without leaving the system in a state your future self will regret.