Senior Full Stack Software Engineer — Commerce Real-Time Video, Computer Vision & 3D

About the role

Senior Full Stack Software Engineer — Commerce - Real-Time Video, Computer Vision & 3D

FormVue · Remote (United States) · Full-time

About FormVue

We are changing the way home solutions are bought and sold. At FormVue, consumers shop with video and AI builds the cart. Our proprietary video stack and AI capture all the information needed to configure, price, and deliver the right products.

We have revolutionized how consumers choose home products and services, and the operational savings and efficiency we drive for manufacturers and dealers are core to the value FormVue delivers.

The role

You own features end to end: browser and mobile capture, the media pipeline, real-time inference, 3D reconstruction, cart and quote generation, CRM sync, and the agents that run the revenue workflow. You ship to production every day. Small PRs, feature flags, trunk-based development, and you watch your code run in prod.

What you'll build

• Browser and mobile video capture that survives bad networks: real-time video streaming, local upload queues and on-device cache, resumable upload, recovery.
• Real-time inference on live streams: frame extraction, GPU inference, results streamed back to the UI in under a second.
• LiDAR/depth capture on iOS (App Clip and full app) and depth capture in the browser on Android; point cloud → mesh → measured 3D model → rendering.
• The real-time data layer: live job status, presence, and streaming updates to homeowners, dealers, and agents.
• Cart and quote generation plus CRM integrations: product configuration, pricing rules, bidirectional sync with systems of record.
• Autonomous agents that qualify, follow up, schedule, and quote, with evals, tracing, guardrails, and human escalation.

Non-negotiables

• You build and ship daily. Show us a live product, a release cadence, a commit history.
• Agentic-native developer. Claude Code and Codex are your primary development environment, not a side tool: agents write, test, and ship code with you every day.
• Practical mindset. You build the simplest thing that works in production, cut scope to ship, and judge your work by customer outcomes.
• Real-time video streaming and video expertise: production WebRTC and HLS, browser and mobile capture, encoding and transcoding, owning the media pipeline end to end.
• Local uploaders and cache: offline-first capture (IndexedDB/OPFS, Service Workers), chunked resumable uploads (S3 multipart, background URLSession), retry, dedupe, background sync.
• Real-time web systems: WebSockets/SSE/WebTransport, pub/sub, event streams, backpressure, idempotency.
• Real-time databases in production: Postgres logical replication/CDC, DynamoDB Streams, Redis Streams, Kafka or Kinesis. Enterprise-grade only.
• Real-time computer vision in production: PyTorch → ONNX/TensorRT, OpenCV, streaming inference against a latency budget; in-browser inference (WebGPU/WASM, ONNX Runtime Web).
• CS fundamentals: data structures, algorithms, concurrency, distributed systems. You can debug threading, memory, and lifecycle bugs under production load.
• Enterprise systems experience: multi-tenant SaaS, SSO/SAML/OIDC, RBAC, audit logs, SLAs, security reviews, PII handling.

Required experience

Agentic-first development

• Claude Code and Codex are your primary workflow. You know their failure modes and structure repos, specs, and tests so agents can work autonomously.
• You've shipped LLM agents that take real actions: tool calling, MCP, structured outputs, multi-agent orchestration, durable workflows (Temporal, Step Functions), evals, tracing, prompt-injection defense, cost and latency budgets.

Ecommerce, CRM, systems of record, revenue automation

• Built or deeply integrated CRMs: Salesforce, HubSpot, ServiceTitan, or similar. Webhooks, bulk APIs, OAuth 2.0, dedupe and identity resolution, lead routing, quote-to-cash.
• Built autonomous agents that act on the CRM (qualify, follow up, schedule, quote) with audit trails and escalation paths.
• Built ecommerce operational systems: catalog and configurable products, cart and checkout, pricing and promotions, payments, order management, fulfillment.
• Systems of record management: ERP/CRM/OMS as source of truth, master data, sync and reconciliation, audit.

LiDAR and 3D

• Point clouds, meshing, SLAM/scene reconstruction, photogrammetry or Gaussian splatting, measurement accuracy. USDZ/glTF pipelines; Three.js or react-three-fiber; WebGPU.
• iOS: ARKit scene reconstruction and depth API, RoomPlan, Object Capture, RealityKit/Metal, Swift/SwiftUI. App Clips: size budget, QR/NFC/link invocation, handoff to the full app.
• Android browser depth capture via WebXR Depth Sensing (ARCore Depth), WebGL/WebGPU point-cloud rendering.

Web app and backend

• Fast, practical front-end work: TypeScript, React/Next.js, React Native, component libraries, performance budgets, polished UI shipped in days, not weeks. Node plus Go or Python on the backend.
• API design at scale: REST and gRPC, GraphQL where it fits, versioning, idempotency keys, rate limiting, backward-compatible contract evolution.
• Data modeling and performance: Postgres/Aurora schema design, indexing and query planning, partitioning, connection pooling, p99 latency work; DynamoDB; knowing when to add a cache, search index, or vector DB.
• Async and distributed systems: SQS/Kafka/Kinesis, outbox pattern, sagas, dead-letter handling, ordering guarantees, retries with jitter, circuit breakers, graceful degradation.
• Reliability and security: SLOs and error budgets, zero-downtime schema migrations, blast-radius containment, OAuth 2.0/OIDC/JWT/mTLS, tenant isolation, secrets management, encryption at rest and in transit.

SysOps / DevOps / platform

• Containers and orchestration: Docker, Kubernetes (EKS) or ECS, autoscaling (HPA/Karpenter), rollouts, resource tuning.
• IaC and CI/CD: Terraform or AWS CDK, Helm, GitHub Actions, blue-green and canary releases, automated rollback.
• AWS operations: VPC/networking, IAM policy design, load balancing, DNS/TLS, cost monitoring and optimization.
• Observability and on-call: OpenTelemetry, Prometheus/Grafana, CloudWatch or Datadog, SLO-based alerting, runbooks, leading incident response and postmortems.
• Resilience and security ops: capacity planning, load and chaos testing, multi-AZ/DR, backup and restore drills, vulnerability scanning, image hardening, SOC 2 evidence.
• Edge: Cloudflare (Workers, Durable Objects, R2, CDN, WAF, Zero Trust), rate limiting, DDoS protection.

Our stack

• Cloud: AWS (ECS/EKS, Lambda, Aurora PostgreSQL, DynamoDB, S3, MSK/Kinesis, ElastiCache, GPU/SageMaker). Cloudflare (Workers, Durable Objects, Containers, R2, Realtime, Stream, Zero Trust).
• Media: real-time video streaming (WebRTC, HLS), Cloudflare Realtime and Stream, FFmpeg/GStreamer; offline-first uploaders and cache.
• Data: Aurora Postgres, DynamoDB Streams, Redis, Kafka/Kinesis, CDC.
• ML/CV: PyTorch, ONNX Runtime, TensorRT, Triton, OpenCV; ONNX Runtime Web/WebGPU in the browser.
• 3D/mobile: Swift/SwiftUI, ARKit/RoomPlan/RealityKit, Metal, Three.js, WebXR, glTF/USDZ.
• Agents: Anthropic/OpenAI/Bedrock APIs, MCP, Temporal, eval and tracing tooling; Claude Code and Codex for development.
• Web: TypeScript, React/Next.js, React Native, Node; Go or Python for services.

This is not a BaaS or no-code shop. You own schemas, streams, and infrastructure directly.

• GPU cost optimization, quantization, edge inference.
• Native Android (Kotlin, ARCore).
• Rust or C++ for media/CV performance work.
• SOC 2, CCPA; treating video of people's homes as sensitive data.

How we work

• Remote, US-based.
• Ship daily. Trunk-based, feature-flagged, you own what you ship (on-call rotation).
• AI-assisted engineering is the default; judgment is required.
• Small team, high ownership, direct access to manufacturers and dealers.

Eligibility

• 5+ years building production software.
• Reside in and be authorized to work in the United States. [Sponsorship: yes/no]
• Fluent, professional English, written and spoken. You'll work directly with manufacturers, dealers, and the team.

Compensation

• Base salary: 150K–220K USD + equity + [benefits: health/dental/vision, 401(k), PTO, home-office stipend].
• CA, CO, NY, WA, and other states require a posted pay range.

How to apply

Skip the cover letter. Send:

• Links to 1–2 shipped products
• GitHub or equivalent.

Process [edit]: 30-min intro → 60-min deep dive on something you shipped → paid 1–2 day practical build → leadership conversation → offer.

Equal opportunity

FormVue is an equal opportunity employer. We do not discriminate on the basis of race, color, religion, sex, sexual orientation, gender identity, national origin, age, disability, veteran status, or any other protected characteristic.