Staff Software Engineer

Quick Facts

• Staff-level hands-on backend engineering role focused on agentic investigations and ML/AI foundations.
• Own backend services and production systems for agent orchestration, retrieval/knowledge grounding, and model integration.
• Design, implement, test, and operate reliable workflows and evaluation pipelines.

Description

You’ll build backend services and AI/ML foundations that power agentic investigation systems. The work covers agent orchestration and asynchronous workflows, dynamic knowledge injection (including structured + unstructured data, knowledge graphs, and hybrid retrieval), model integration, and evaluation with strong production observability. You will also help agents preserve evidence and provenance so downstream decisions are accurate and reliable.

Responsibilities

• Design, implement, test, and operate backend services for agent orchestration, task execution, state management, model integrations, and investigation workflows.
• Build synchronous and asynchronous agent task workflows (queues, retries, resumability, cancellation, idempotency, scheduling, failure recovery).
• Contribute to agent frameworks and developer tooling for tool calling, structured outputs, reusable skills, actor-based workflows, human review, and policy-controlled execution.
• Implement dynamic knowledge injection using investigation type, entity context, workflow state, customer policy, prior findings, and task requirements.
• Develop knowledge-grounding capabilities distinguishing authoritative facts, policies, derived signals, hypotheses, and unverified observations.
• Build knowledge ingestion/enrichment pipelines (normalize, classify, chunk, index, version, refresh structured and unstructured data).
• Implement hybrid retrieval (semantic + lexical, metadata filtering, reranking, entity-aware, temporal filtering) and multi-stage retrieval.
• Contribute to knowledge-graph systems (entities, relationships, events, evidence, provenance, risk signals) and graph-enhanced context assembly.
• Add context-selection logic balancing relevance, freshness, authority, latency, tenant permissions, and model context limits; prevent stale/contradictory/poisoned/unauthorized context.
• Implement agent memory/state (task state, investigation history, entity context, retrieval state, provenance).
• Integrate agents with internal APIs, identity and risk sources, case management, watchlists, model endpoints, webhooks, and event streams.
• Build secure connectors/tool execution patterns (authentication, authorization, credential management, tenant isolation, audit logging, sandboxing).
• Integrate ML models for classification/ranking/extraction/entity resolution/summarization.
• Partner on ML pipelines: feature definitions, labeling, dataset quality, model interfaces, and deployment workflows.
• Contribute to model training/fine-tuning/adaptation and evaluation/versioning/deployment/rollback.
• Implement model serving/routing to select models by task needs (latency/cost/quality/context/data boundaries).
• Improve performance with caching, batching, parallel execution, retrieval optimization, context reduction, prompt/model selection, and efficient inference.
• Build evaluation/regression tests for retrieval quality, grounding/citations, tool selection, workflow completion, and end-to-end outcomes.
• Instrument services (logs, metrics, traces, latency, cost attribution, retrieval diagnostics, model version tracking, dashboards) and use telemetry + feedback to improve systems.
• Participate in on-call and incident response; improve monitoring, SLOs, capacity planning, and recovery.
• Write technical designs and document decisions; communicate trade-offs; collaborate across Product, Fraud/Risk, Investigation Ops, Platform/SRE/Security/Privacy, and Data Science.
• Contribute to engineering standards for testing, observability, deployment, security, maintainability, and responsible AI.

Benefits

Not specified in the provided posting.