AI Engineer

Overview HubSyncis an end-to-end platformfor tax andauditfirms.Wehandle document management, workflow automation, and AI-powered document processing for CPA firms during their busiest periods. We aretransformingHubSyncinto an AI-native platform withagentic AI capabilities acrossall our key modules. We are looking foran experienced backendengineer who has built and shipped production systems that real users depend on.

What We Are Building Active areas of work where you will have direct ownership: Agentic workflow orchestration.Multi-agent coordination across tax document workflows with human-in-the-loop oversight. Agent state machines, tool routing, context windowing, and retry semantics for processes that run for minutes or hours. Document intelligence atscale.Production-grade pipelines that extract, classify, andvalidatetax forms and financial documents across dozens of formats and quality levels.

Workflow state management.State hydration for long-running agentic workflows, failure handling, checkpoint/resume, and recovery across distributed services. Evaluation and observability.Task completion rates, accuracy attribution, cost tracking per action, regression detection. Attributing outcomes to specific agent reasoning steps when something goes wrong.

Cost-accuracy optimization.Optimizingcost, accuracy, and latency trade-offs across different document types, complexity levels, and client tiers during peak tax season volume. Trust and reliability.Making non-deterministic agent outputtrustworthyfor professionals who cannot accept errors. Supervision layers, validation rules, humanreview gates.

What We Look For You have built and shipped production systems.You have taken features from design through implementation to production release, and you have kept them running.You understand concurrency, failure modes, data integrity, and why things break at scaleand can architect solutions end-to-end. You can point to enterprise-grade features and products that youbuiltand that users rely on today. You have dealt with the full lifecycle: requirements, implementation, testing, deployment, monitoring, and theproductionincidents that follow.

You work across the stack when the problem requires it. The boundaries between backend, data, infrastructure, and product work are not rigid here. The best work happens when engineers move between them based on what the problem demands.

Must Have 4+ years building and shipping backend systems in production environments where uptime and correctness matter A track recordof delivering enterprise-grade features and products, from design through deployment and ongoing operations Deep experience with relational databases: PostgreSQL or equivalent, schema design, query optimization, data modeling, migrations Hands-on work with event-driven architectures: message queues, async processing, distributed job execution Production experience with AWS (Lambda, SQS, S3, ECS) or equivalent cloud platforms Comfort reading and writing both TypeScript and Python (or the ability and willingness to pick up a second language quickly) Experience with the full software delivery lifecycle: design, implementation, testing, deployment, monitoring, and incident response Good to Have Exposure to agentic systems, agent orchestration frameworks, or multi-agent workflow design Familiarity with RAG architectures, vector databases, or document processing pipelines Experience with multi-tenant SaaS architecture (schema isolation, tenant-scoped data, access control) Background in document intelligence: OCR, structured extraction from PDFs, form understanding Open-source contributions, technical writing, or other public evidence of engineering depth Technologies & Frameworks Languages & Runtimes TypeScript Python 3.12 React 18 with module federation formicrofrontendarchitecture Node.js 20 (Fastifyand Express services) AI & Agent Infrastructure AWS Bedrock,AgentCore(Claude, Titan embeddings, cross-encoder reranking) LangGraphfor agent orchestration and state machine management LangChainfor tool chaining and model integration MCP (Model Context Protocol) for dynamic tool generation fromOpenAPIspecs #J-18808-Ljbffr