Principal Software Engineer — Platform, Pipelines & Security

Diopter is a real-time defense for people in live conversations. We watch phone calls and video meetings as they happen, with email coming next, for a cloned voice, a fake face, and the pressure an attacker applies to rush someone into a decision. Instead of checking whether one clip is fake, we follow the whole conversation and compare where it is heading against the rules a company already has for approving money, hiring, and access. When something is off, the person on the call gets a clear warning and a next step while there is still time to stop. We are building this now because AI is about to let a small number of attackers run these conversations at a volume no human team can keep pace with, and the human layer is the part of a company nobody has protected yet. **About the Role** We're hiring a Principal Software Engineer to build the foundation that the rest of Diopter's technology runs on. You'll design and ship the high-throughput media streaming pipelines, the backend services, and the customer-facing frontend that together make up our platform. You'll also make sure it deploys cleanly and securely into our cloud and into customer data centers.This is a hands-on, full-stack principal role. You'll set the architecture, write a lot of the code, and hold the bar for how everyone else builds on top of it. Since we're a security company, you'll treat every design decision as a security decision. **What You'll Do** * Own the core platform architecture, from design through rollout to long-term operation: * Real-time media pipelines: ingest, transport, chunking, buffering, and fan-out of live audio and video streams (meetings, calls, uploads) with predictable latency and backpressure under load. * Backend services: high-concurrency APIs and processing services in Go and Python, including job orchestration, queueing, deadlines, retries, and idempotency. * Frontend: the customer dashboard, admin console, and integration surfaces, built in TypeScript and React to be fast, clear, and secure by default. * Make on-prem a first-class deployment target. Design one platform that runs on AWS, in customer data centers, in private clouds, and in air-gapped environments. This includes packaging (Helm, OCI images), offline install and upgrade paths, hardware sizing (including GPU nodes), and supportability when we can't SSH in. * Build for scale and reliability. Set SLOs, capacity models, autoscaling strategy, failure isolation, and graceful degradation for pipelines that cannot drop or reorder customer media. * Drive security across the stack. This covers authN/Z (OIDC, OAuth, SSO, SCIM), multi-tenant isolation, encryption in transit and at rest, secrets and key lifecycle, audit logging, least-privilege infrastructure, and supply-chain integrity (signed images, SBOMs, dependency scanning). * Threat-model everything. Lead design reviews and threat models for major initiatives. Treat media inputs, customer integrations, and every network boundary as hostile, and drive systemic issues to closure. * Be the technical counterpart for B2B customers. Work with enterprise security and IT teams on deployment architecture, network and firewall requirements, SSO integration, and security questionnaires and reviews (SOC 2, ISO 27001). * Lead cross-functional delivery with engineering, research, product, and customer teams. Shape interfaces, migration plans, and safe rollouts across dev, QA, staging, production, and customer sites. * Mentor engineers and raise the bar on code quality, operational readiness, observability, and secure-by-default practice. **Our Stack** * **Backend:** Go, Python, PostgreSQL, Redis, S3-compatible object storage * **Frontend:** TypeScript, React, modern web tooling * **Media:** WebRTC, Web Audio, real-time audio/video processing, browser-automation-based meeting integrations * **Compute:** Kubernetes (EKS and on-prem), Karpenter, NVIDIA GPU nodes, Helm * **Delivery:** GitLab CI, ArgoCD (GitOps), OpenTofu, multi-account container registries * **Identity & Secrets:** Zitadel (OIDC/OAuth), OpenBao, External Secrets Operator, IRSA, SSO * **Observability:** Structured logging, metrics, tracing, in-house Go observability library **Requirements** * 8+ years building and operating production distributed systems, including ownership of on-call and incident response. * Proven experience building high-throughput streaming or media pipelines, such as real-time audio/video, event streaming, or large-scale data processing, with a clear understanding of latency, backpressure, and failure modes. * Strong full-stack skills. You're deep in the backend (Go and/or Python) and capable on the frontend (TypeScript/React), and you can own a feature end-to-end. * Hands-on on-prem and hybrid deployment experience. You've shipped software that customers install in their own data centers or air-gapped environments, including packaging, upgrades, licensing, and remote supportability. * Deep Kubernetes and cloud expertise (AWS required, others a plus), including networking, IAM, storage, and multi-environment design. * A cybersecurity mindset. You have strong fundamentals in authN/Z, cryptography in practice, network security, tenant isolation, secure SDLC, and common vulnerability classes (OWASP, cloud misconfiguration, and supply chain). You think like an attacker when you design. * B2B and enterprise experience. You're comfortable in front of a customer's CISO or infrastructure team and can answer their hardest deployment and security questions. * The ability to lead cross-team technical initiatives: set direction, align stakeholders, deliver measurable outcomes, and write design docs people actually read. * Clear communication with engineers, executives, and customers. Nice to Have * Experience with WebRTC, SIP, media servers, or codecs. * Experience running GPU workloads on Kubernetes, on-prem, or in the cloud. * Background in security products such as SIEM, SOAR, EDR, identity, or fraud prevention, and in integrating with them. * Experience with compliance frameworks (SOC 2, ISO 27001, FedRAMP) or regulated industries (finance, government, healthcare). * Familiarity with using ML models in production pipelines. * Early-stage company experience, where "principal" means building the foundation, not only reviewing it.