You own a system, not a ticket queue
Roles here are defined by what you're responsible for in production, not by how many story points you close. You'll be the person who knows why the architecture is the way it is.
Careers
Work at Talvex
We're a small, senior team. Every person here owns a system end to end and talks to the client who depends on it — there is no layer between you and the people using what you build.
What working here actually means
You talk to clients directly
Engineers and designers join the client calls for their own work. No account manager relays requirements, and no requirement reaches you third-hand.
Senior means senior
We hire few people and pay for judgment. If your role says ten years, you'll be the most experienced person in the room on that subject and we'll defer to you on it.
Written decisions, async by default
Architecture decisions, trade-offs, and scope changes live in documents. Meetings are for the things documents genuinely can't settle.
Open roles
All roles are remote-first and full-time. If you're close but not an exact match, apply anyway and say where the gap is — we'd rather read that than a rewritten CV.
- 10+ years
- Full-time · Remote
Full-time · Remote
We build and modernize products for US health technology companies, which means systems holding protected health information under HIPAA. This role owns how those systems are designed to be defensible: the threat model, the controls, the evidence, and the answers when a client's enterprise customer sends a 300-line security questionnaire.
What you'll own
- Security architecture for client platforms processing PHI, from design review through production
- Threat models and data-flow diagrams that stay current as the system changes
- HIPAA Security Rule control mapping, and the technical evidence behind each control
- Cloud security posture on AWS and GCP: identity, network segmentation, key management, audit logging
- Incident response runbooks, and leading the technical side when one is used
- Security review of our own delivery pipeline — dependencies, secrets, access, and release gates
- The technical portion of client security questionnaires, vendor assessments, and audit support
What we look for
- Ten or more years in security engineering or architecture, with several spent on systems holding regulated health data
- Direct working knowledge of the HIPAA Security Rule and what auditors actually accept as evidence
- Experience taking an organization through SOC 2 Type II or HITRUST, on the technical side rather than the paperwork side
- Depth in cloud identity and key management — you can review an IAM policy and say precisely what it grants
- The ability to write a control decision down clearly enough that an engineer implements it correctly without you in the room
- Comfort telling a client that something they want to ship is not defensible, and proposing what is
Nice to have
- US healthcare integration experience: HL7 v2, FHIR, or payer/claims data
- Application security depth — code review, SAST/DAST tuning, dependency risk
- Kubernetes security: admission control, workload identity, network policy
- A CISSP, CCSP, or equivalent, if you have one — we don't require it
- AWS
- GCP
- Terraform
- Kubernetes
- HIPAA
- SOC 2
- HITRUST
- OpenTelemetry
- 5+ years
- Full-time · Remote
Full-time · Remote
Several of our platforms live or die on whether people return to them week after week — a sports platform with players and organizers, SaaS products with daily operational users. This role designs the engagement systems behind that: progression, status, streaks, challenges, and the notifications that carry them, then proves with data whether they worked.
What you'll own
- Engagement and retention mechanics for client products: progression, achievements, streaks, leaderboards, seasons
- The behavioural model behind each mechanic — what it rewards, who it excludes, and how it fails
- Lifecycle and notification design across in-app, email, and push, including how often is too often
- The experiment plan: hypothesis, metric, sample, and a defined stopping point before the test runs
- Onboarding flows measured by activation, not by completion
- Post-launch reporting to the client on what moved, what didn't, and what you recommend next
What we look for
- Five or more years designing engagement, retention, or loyalty systems for a live product with real users
- Evidence you have shipped mechanics and measured them — cohort retention, activation, repeat rate, not vanity counts
- Fluency in the analytics side: you can define events, read a funnel, and spot when a result is noise
- Judgment about incentives — you can name the mechanics you have chosen not to ship, and why
- Clear writing, since much of this role is convincing a client that a behavioural change is worth building
Nice to have
- Community, sports, marketplace, or education product experience
- Working knowledge of SQL, enough to answer your own questions
- Experience with A/B testing infrastructure and its statistics
- Some background in behavioural science or game design
- Amplitude / PostHog
- SQL
- Braze / Customer.io
- Figma
- Notion
- 4+ years
- Full-time · Remote
Full-time · Remote
AI features break most of the assumptions interface design rests on: responses take seconds, arrive in fragments, and are occasionally confidently incorrect. This role designs for that reality — the waiting, the uncertainty, the correction, the moment a person decides whether to trust the output — across the AI products we build for clients.
What you'll own
- End-to-end design of AI-facing surfaces: assistants, generation flows, review and approval interfaces
- The states most teams skip — streaming, partial output, low confidence, refusal, and recovery from a wrong answer
- How a user corrects the system, and what the product does with that correction
- Design of the human review step in agent workflows, so the reviewer can actually judge quickly
- Prototypes real enough to test with live model output, not static screens with invented text
- The design system for each product, kept small enough that engineers stay inside it
What we look for
- Four or more years in product design, with at least one shipped AI or ML-driven feature you can walk us through
- A portfolio that shows reasoning, not just final screens — what you tried, what failed, what you cut
- Real fluency with AI tooling in your own process, used for speed rather than as a talking point
- Enough comfort with the technical side to know why a response streams and what that costs the interface
- Strong writing — in AI products the copy is the interface
Nice to have
- Prototyping in code, even roughly: React, Tailwind, or the equivalent
- Experience designing for evaluation or annotation workflows
- Data-dense B2B interfaces: tables, filters, bulk actions
- Accessibility work you have done deliberately rather than as a checklist
- Figma
- React
- Tailwind CSS
- Claude
- Framer
- 5+ years
- Full-time · Remote
Full-time · Remote
This role owns the infrastructure under client products on GCP: how it's provisioned, how it deploys, how it's observed, and what it costs. Not a ticket-driven ops role — you'll design the platform, then be the person on the call when production has a bad night.
What you'll own
- GCP infrastructure defined entirely in Terraform, with reviewed changes and no console drift
- GKE clusters: workload identity, autoscaling, network policy, upgrade strategy
- CI/CD pipelines with staged rollouts and a rollback that has been rehearsed rather than assumed
- Observability — metrics, logs, traces, and alerts that fire on user impact instead of CPU graphs
- Cloud SQL and data-layer operations: backups, restore drills, connection management, failover
- Cost: visibility per environment, and the changes that bring the bill down without hurting reliability
- Production incidents end to end, and the written post-mortem afterwards
What we look for
- Five or more years in DevOps, SRE, or platform engineering, with GCP as your primary cloud in production
- Terraform at a level where you have handled state migrations and module design, not just applied plans
- Kubernetes in production, including the failures — you have debugged a cluster that was actively breaking
- Strong Linux and networking fundamentals, and enough Go or Python to build your own tooling
- Experience carrying an on-call rotation, and clear views on what makes an alert worth waking someone
- The instinct to automate a task the second time you do it, and to document it the first
Nice to have
- Multi-cloud experience, particularly AWS alongside GCP
- Platform hardening for regulated workloads
- Data pipeline operations: Dataflow, Pub/Sub, BigQuery
- A GCP Professional certification, if you have one — we don't require it
- GCP
- GKE
- Terraform
- Cloud SQL
- Cloud Build
- GitHub Actions
- Prometheus
- Grafana
- Go
How hiring works
Four steps, roughly two weeks end to end. You'll always know what the next one is and when to expect it.
- 1
Application review
A written answer within five business days — including when the answer is no. If we pass, we tell you the actual reason.
- 2
Intro call — 45 minutes
A conversation with a founder about your work and what the role really involves. You'll get honest answers about the projects, the clients, and the money before you invest more time.
- 3
Technical conversation — 90 minutes
A discussion of real problems from our work, in your area. No whiteboard algorithms, no take-home task built on unpaid work you'll never hear about again.
- 4
Offer
A written offer with the compensation, the engagement terms, and the first project named. We don't negotiate by making you guess a number first.
Apply
Send us the role you're after, a short note on what you've owned, and anything that shows your work — a repository, a portfolio, a system you designed. A CV is fine; a rewritten one to match the posting isn't necessary.
Nothing here fits but you think we should talk? Send it anyway and say what you'd want to own.
No agencies, and nothing generated wholesale by a model — we read every application ourselves and it shows.