Infrastructure Engineer
Austin, Texas, United States, New York City, New York, United States
About Us
Insurance brokers sit at the intersection of care, cost, and access — yet remain one of the most
underleveraged assets in the $5T healthcare value chain. They guide millions of Americans through
plan selection, coverage questions, and care navigation, yet still rely on spreadsheets, manual
workflows, and outdated tools that limit their growth and impact.
Gyde is reimagining this. We’re building the first AI-native insurance brokerage platform—a system that
learns from every client interaction to automate operations, power intelligent voice and chat
experiences, and predict the right coverage and products for every individual or business.
Our approach combines acquisition and AI: we acquire traditional brokerages and transform them into
next-generation, data-driven organizations. Through Gyde’s platform, agencies run more efficiently,
serve clients more personally, and scale faster than ever before.
Join us if you’re excited to:
Redefine how millions of people access and understand their healthcare coverage, by building
systems that turn unstructured data, human conversations, and fragmented processes into
intelligence.
Build AI systems that improve how people access healthcare
Design production-grade voice, chat, and predictive models for a highly regulated domain
Launch technologies that augment human judgment and make complex decisions transparent
and scalable
Help reinvent an entire industry from the inside out
Our founding team boasts pedigrees from Oscar, Stripe, Vista Equity Partners, Bain & Company and
Alpine. Lightspeed led Gyde’s $60M financing, with participation from Optum Ventures, Crystal Venture
Partners, Virtue, MVP Ventures, and multiple endowment funds, among others. Hear why on LSVP
Investment Memo.
Role Summary
You own the infrastructure that Gyde’s AI-native brokerage platform runs on — but at our size, the title
undersells the job. We acquire insurance agencies and fold them onto one platform, which means our
environment grows in scale and in tenancy every quarter, and it has to stay secure, auditable, and
predictable while it does. You sit on the Infrastructure team and report to our Head of Infrastructure.
The work deliberately blurs the usual line between infrastructure and application engineering. The cloud
layer — EKS, AWS in Pulumi, CI/CD, the observability stack — is table stakes, but it is rarely the whole
deliverable. A typical project runs the full chain: a business problem lands (“we can’t tell which agency
this contract belongs to”, “we can’t tell whether that email arrived”, “we don’t know what we’re spending
on LLMs”), and you own it from audit and data model through to the TypeScript that ships in a shared
package, the Temporal workflow that runs it, the IAM role behind it, and the dashboard that proves it
works.
So this is real application engineering as well as platform work — TypeScript in our monorepo, Temporal
workers, Postgres and SQL migrations, packages other engineers depend on. A third part of the job is
neither infrastructure nor application code: evaluating a vendor’s permission model closely enough to
know what it can actually enforce, designing an approval workflow with named humans and SLAs, and
treating HIPAA and cost as design inputs rather than review gates. Because we’re small, you are often
the interface to security, data, finance, and legal at once. You will write the decision doc, then defend it
in a comment thread with someone who cares about the price rather than the architecture.
The throughline is simplification. Anyone can add another tool. We want someone who takes a complex,
regulated, multi-tenant environment and makes it feel simple to the people using it — and who knows
that the mechanism is usually the short part, while the carve-outs, the sequencing, and the question of
whether a control is truly enforced are the job.
Key Responsibilities
Own the chain from problem to running control — take an ambiguous business or platform
problem, establish what is actually true today rather than what the docs claim, design the fix,
build it, and instrument it so you can prove it works. Most projects here cross the infra/app line
and you own both sides.
Build in the monorepo — production TypeScript in shared packages other engineers import,
Temporal workers and scheduled workflows, data models and SQL migrations, internal services
and tooling. Not just YAML wrapped around someone else’s code.
Run the cloud platform — EKS, AWS defined in Pulumi, CI/CD, secrets and workload identity, and
the metrics, logging, and alerting that tell us something is wrong before a broker does.
Evaluate and integrate the systems we buy — identity, orchestration, model gateways,
warehouse and vendor APIs. Read a permission model and plan tier closely enough to know
what a tool enforces versus what a person is merely promising to do, then build the glue that
makes them agree.
Make security and compliance enforceable — least-privilege IAM and workload identity, BAA
before PHI access, data classification, PHI-aware logs and metric labels, and audit trails that
hold up in a carrier audit because the system produces them, not because someone
remembered.
Own cost as an engineering surface — per-key LLM and infrastructure spend,
estimated-versus-billed reconciliation, and alerting that fires before a runaway job becomes an
invoice.
What you bring and who you are
Cloud & Kubernetes Foundations — Production EKS and AWS expressed in code, CI/CD you
would stake a release on, and observability people actually act on. You have operated clusters
through upgrades and incidents, not just deployed to them. Here this is table stakes rather than
the whole job.
Application Engineering in a Shared Codebase — You write real TypeScript, not only glue.
Durable workflows, relational data modeling, migrations, and packages other engineers import.
You are comfortable opening a pull request in application code you do not own, and you know
when a shared abstraction earns its keep.
Integration & Vendor Systems — You reverse-engineer an admin console rather than trusting
its marketing page. You check plan tiers and API surfaces before committing to a design, and
you can tell the difference between a control a system enforces and a process a human
promises to follow.
Security, Compliance & Cost as Design Inputs — Least privilege is a default, not a cleanup task.
You know where PHI leaks in a stack — logs, traces, metric labels, backups, third-party payloads
— and you treat spend as something you instrument, not something finance discovers on an
invoice.
Judgment, Sequencing & Written Decisions — You verify the premise before you build. You
think in blast radius, carve-outs, and reversible-first rollout, because the first casualty of
enforcing something early is usually a business process. You write the one-page decision doc
with ranked options and open questions, and you can defend it to a non-engineer.
Technical Requirements
5–10 years in infrastructure, platform, SRE, or DevOps engineering, with meaningful production
ownership.
Production Kubernetes experience — EKS preferred: cluster upgrades, autoscaling, networking,
and hands-on troubleshooting.
Strong AWS fundamentals: IAM, VPC and networking, compute, storage, KMS, and a working
grasp of the cost model.
Infrastructure as code in a real codebase — Pulumi preferred; Terraform or CDK acceptable with
a willingness to work in Pulumi.
Production TypeScript or Node — you will ship application code in a shared monorepo, not only
infrastructure definitions. We care that you have shipped real application code and will work
primarily in TypeScript; which language you did it in matters less.
Relational data modeling and SQL, Postgres preferred; comfortable writing and reviewing
migrations.
Built and owned CI/CD pipelines (GitHub Actions or comparable), including deployment
strategy and rollback.
Observability in practice — metrics, logs, and distributed tracing, plus alerting that engineers
trust.
Experience operating in a HIPAA, SOC 2, or otherwise regulated and audited environment.
A writing habit — design docs, RFCs, or postmortems you can point to.
Fluency with AI coding tools in day-to-day work — we are an AI-native company and expect
engineers to use agents well.
Nice to have
Durable workflow engines — Temporal especially — or comparable orchestration and job
systems.
Hands-on administration of an IdP or SaaS admin surface (Okta, Entra) including SCIM, RBAC,
and API integration.
Multi-tenant SaaS, or M&A integration work folding acquired companies onto a single platform.
Data warehouse pipelines (Redshift, BigQuery, Snowflake) and event-ingestion plumbing.
LLM infrastructure: gateways and virtual keys, provider rate limits, token and spend telemetry.
Secrets management at scale, policy-as-code, and supply chain hygiene (image signing,
SBOMs).
What we offer
Gyde offers a competitive benefits package to all employees.
Flexible (Unlimited) Paid Time Off
Hybrid Work in Austin or NYC
Medical, Dental, and Vision benefits for you and your family
Retirement Plan (e.g., 401K)
Parental Leave