Senior Platform Engineer / SRE

Senior Platform Engineer / SRE

About the role

Tangible runs a liquidity platform for private-market assets, used by some of the largest wealth and asset managers in the world. Each major partner gets its own production environment on AWS, and the platform is regulated, audited and integrated with their systems.

How we build software has changed. Much of our code is now written by AI agents working in parallel, reviewed by engineers and shipped through pull requests. We also run AI in production: agents that do research, MCP servers that let people query our platform in plain language, and LLM calls inside product workflows.

We need a Senior Platform Engineer / SRE to own the foundation for all of this. You'll keep our multi-tenant AWS estate reliable, secure and cheap to run. You'll also build the guardrails that let both humans and agents ship quickly without breaking production. This is an engineering role, not a ticket queue. You'll write Python and TypeScript, and the platform itself is your product.

What you'll do

Reliability and infrastructure

- Own Tangible's AWS estate end to end (ECS Fargate, RDS Postgres, Redis, S3, ALB, KMS, Lambda), with a separate environment for each client tenant

- Own uptime: define SLIs and SLOs per tenant, set error budgets, and build alerting that pages on customer impact rather than noise

- Lead incident response, blameless postmortems and long-term fixes, including the uncomfortable ones such as undersized databases and migrations that take locks

- Keep everything in Terraform: reproducible environments, drift detection, and a way to create a new tenant environment in hours instead of weeks

- Make costs visible and keep them down, broken out per tenant and per service, LLM token spend included

A platform built for humans and agents

- Design environments where AI coding agents can work safely: sandboxed, isolated preview environments that start up quickly; agents only change code through PRs; credentials scoped to what each agent needs and short-lived

- Turn standards into automated checks in CI (policy as code, security scanning, schema-change gates, contract checks), because reviewers can't keep up with how much agents produce

- Write runbooks and operational knowledge in a form both an engineer and an agent can follow, so agents can do first-pass incident triage and a human makes the call

- Build internal tools and services in Python and Node.js/TypeScript that remove manual work for the engineering team

Running AI in production

- Host and operate our AI workloads: agent workers, MCP servers behind SSO, scheduled agent jobs

- Run a model gateway: one place to manage provider keys, rate limits, fallbacks, logging and spend per tenant

- Put the controls in place that a regulated environment needs: audit trails for agent actions, boundaries on data and PII, defences against prompt injection and data exfiltration, and a clear record of which model touched what

Security and compliance

- Improve security and operational maturity across all systems, with SOC 2 controls built into the platform instead of bolted on afterwards

- Manage identity and access for people and machines (SSO/OIDC, IAM, secrets management) so nobody has broad standing access

What we're looking for

- 5+ years in platform engineering, SRE or infrastructure roles, including production systems

- Deep hands-on AWS experience: compute, networking, IAM, RDS, observability

- Strong Terraform (or similar) experience, and CI/CD with GitHub Action

- Production-grade Python and/or TypeScript/Node.js. You build services, not just glue scripts

- Solid Postgres operational knowledge: performance, locking, backups and safe migrations

- Real experience with observability (metrics, logs, traces) and running incidents

- You already work daily with AI coding agents (Claude Code, Cursor, Codutocomplete. You can show how you've changed your workflow, what you hand to agents, what you don't, and how you check their output

- A clear view of the risks AI brings to infrastructure: over-privileges, leaked secrets, prompt injection, runaway token costs

- Clear written English. A lot of our coordination is asynchronous and in writing

- Comfortable with startup ownership: you stay calm in a production incident afterwards

Nice to have

- Experience in fintech or another regulated industry, or with single-tenant deployments for enterprise clients

- Experience running LLM workloads in production (model gateways, evals,

- Experience building or hosting MCP servers or agent runtimes

- Containers beyond ECS (Kubernetes, Fly, Coolify), Cloudflare

How you think                                                                                                                                                   You'd rather build the guardrail than act as one. You know a one-person team can scale if the platform does most of the work. You treat agents as a big productivity gain and as untrusted contributors at the same time, and you design for both. When something breaks you fix it and make sure that kind of failure can't happen again.