Senior Software Engineer, Python + AI Platform

Job description

**Who are we?** Smarsh empowers its customers to manage risk and unleash intelligence in their digital communications. Our growing community of over 6500 organizations in regulated industries counts on Smarsh every day to help them spot compliance, legal or reputational risks in 80+ communication channels before those risks become regulatory fines or headlines. Relentless innovation has fueled our journey to consistent leadership recognition from analysts like Gartner and Forrester, and our sustained, aggressive growth has landed Smarsh in the annual Inc. 5000 list of fastest-growing American companies since 2008. Smarsh is hiring a senior backend/platform engineer to build and scale agentic AI systems for enterprise use. You will build fast-moving, early-stage Python services that integrate AI capabilities into a production agentic platform. Your scope spans workflow execution, scale, reliability, and platform hardening as we grow. This is not a generic backend role. The focus is building and designing agentic systems, shipping working software, and solving hard platform problems in a fast-moving AI-native environment. You will join a small, high-velocity cross-functional group and own problems end to end, designing and building from scratch, making fast architectural calls, and driving ideas from whiteboard to working system with a small, high-agency team. ### What will you do? - Drive backend development for AI workflows as part of a collaborative team. Build and evolve Python/FastAPI services powering core agentic workflows and platform capabilities. - Productionize LLM integrations. Implement systems around Bedrock usage, quotas, retries, failover, cost controls, model configuration, and approval constraints. - Design for security and compliance. Address customer data handling, tenant isolation, auditability, observability, and secure processing for regulated workloads. Apply auditable data design patterns to ensure AI outputs are traceable, reproducible, and built to withstand regulatory scrutiny. - Build for scale. We're a nimble team, but our enterprise customers process data at petabyte scale. Help the platform grow to meet that bar through async job orchestration, performance tuning, and data-layer optimization. - Support multi-tenant architecture. Contribute to tenant-aware services, role-based access, SSO integration, and admin/reporting capabilities. - Improve platform reliability. Add monitoring, tracing, alerting, and operational tooling for LLM pipelines, workflow execution, and report generation. - Build real-time capabilities. Design and implement real-time event delivery and pub/sub patterns to support live workflow state, notifications, and agent feedback loops. - Contribute to technical decisions. Partner on shared services decisions, platform architecture, and integration boundaries across the stack. - Work across ambiguity. Translate evolving product requirements and non-functional requirements into practical technical solutions with product, architecture, legal, and security stakeholders. - Champion code quality. Drive strong typing, automated testing, and continuous integration practices that keep the team fast and safe. - Design typed API contracts. Own the API surface as a product contract: designing clean, schema-driven APIs that support typed client generation and reliable integration across services. ### What will you bring? - Strong Python backend engineering. 7+ years professional software development, including 5+ years building Python services in production. Deep experience with APIs, async processing, background jobs, and workflow orchestration. - Cloud-native backend experience. AWS experience, ideally with services relevant to secure enterprise workloads (compute, storage, networking, CI/CD, identity, secrets, encryption). - Production distributed systems. Proven ability to productionize complex backend systems with reliability, observability, retries, throughput, failure handling, and performance tuning. - Data-intensive system design. Strong knowledge of PostgreSQL, large-scale data processing patterns, indexing, query tuning, and batch/stream tradeoffs. Experience with retrieval-augmented generation (RAG), vector search, and embedding-based systems is required (not a plus). - Security and compliance mindset. Experience with multi-tenant systems, RBAC, audit logging, secure data handling, and regulated environments. - Strong ambiguity handling. Ability to work from partial requirements and shape implementation around product and non-functional requirement constraints. - Agentic workflow engineering. Hands-on experience building LLM-driven workflows: tool-calling, state machines, human-in-the-loop approval patterns, checkpoint/resume, and multi-step agent orchestration. Familiarity with frameworks like LangGraph or equivalent. - AI-native engineering. Experience working on or alongside AI-native engineering teams, where AI agents are first-class participants in the development workflow, not just productivity tools. Includes hands-on prompt engineering, eval design, and LLM cost optimization: caching strategies, token efficiency, and model selection tradeoffs. - Product mindset. Bias for shipping, learning from real usage, and making pragmatic tradeoffs grounded in customer problems. ## Strong Pluses - LLM / AI platform experience. Bedrock, OpenAI, Anthropic, LangChain/LangGraph, prompt workflows, evals, tool-calling systems. Experience integrating external AI services safely and reliably. - Identity and access. SSO/SAML/OIDC, enterprise auth patterns. - Graph-shaped data and entity resolution. Experience with graph-backed data models, entity deduplication, mention linking, and building systems that reason over connected, structured records. - Observability stack. OpenTelemetry, tracing, metrics, alerting, cost/usage dashboards. - Regulated communications or compliance domain. Background in systems that handle sensitive communications, audit trails, or data subject to legal or regulatory review is a meaningful differentiator. - Infrastructure as code. Terraform, feature flags, canary deployments, release strategies.