In so far as I deploy on a Linux host (without a container abstraction), I prefer working in Fedora and RHEL environments to Debian-based environments.
I like the adherence to upstream, strong free software stance, and good documentation from Fedora. Clients enjoy cutting edge software availability and a safe fallback to Red Hat Enterprise Linux or CentOS.
With a little forethought, ECS can handle a good portion of my development stack as though it were production. 12 Factor configuration makes this a breeze.
This isn't exactly low-latency (10s to 100s of milliseconds), but it has good throughput and a simple API. There is good reliability, and there is no configuration necessary to get up and running. A hosted queue is important when trying to move fast.
S3 is great for backups, data blobs, as well as log and event streams using Factual's s3-journal. s3-journal really is a secret weapon.
Redis makes certain operations very easy. When I need a high-availability store, I typically look elsewhere, but for rapid development with the ability to land on your feet in prod, Redis is great. The available data types make it easy to build non-trivial indexes that would require complex queries in postgres.
Rapid stand up of related services for testing while enabling artifact creation of services for promotion to prod.
Relational data stores solve a lot of problems reasonably well. Postgres has some data types that are really handy such as spatial, json, and a plethora of useful dates and integers. It has good availability of indexing solutions, and is well-supported for both custom modifications as well as hosting options (I like Amazon's Postgres for RDS). I use HoneySQL for Clojure as a composable AST that translates reliably to SQL. I typically use JDBC on Clojure, usually via org.clojure/java.jdbc.
Elasticsearch has good tooling and supports a large api that makes it ideal for denormalizing data. It has a simple to use aggregations api that tends to encompass most of what I need a BI tool to do, especially in the early going (when paired with Kibana). It's also handy when you just want to search some text.
I mostly use React via a Clojure wrapper called Reagent. Reagent is dead simple and faster than bare React is out of the box. If I have to write a UI myself, I would strongly move to use React via Reagent.
Cloure is a high level language that provides access to both the JVM (for server-side development) and javascript (for client-side development) with largely the same language. This is important to limit context switching and enable code-reuse during fast product cycles. Clojure is ideal for rapid prototyping and has a strong focus on stability, correctness, and concurrency. Tools like Schema and Spec enable well-structured development and high code confidence.