We choose the stack by load profile, team maturity, change velocity, and support cost rather than by fashion.
02
/
stack
Technology Stack
We're not tied to a single framework. Every project gets the stack that's optimal for its requirements, load profile, and team. But we have proven tools we come back to.
For new products we keep a clear baseline set of decisions that enables a fast start without architectural improvisation.
If a project already lives on another stack, we do not break it for aesthetic reasons: first we price the migration, then we discuss changing the contour.
WEB
Frontend
SPA, SSR, ISR — we choose the approach based on the task. Our standard stack for data-heavy dashboards and public products.
primary tools
React 19
foundation for SPA and SSR apps
Next.js 15
App Router, RSC, ISR
TypeScript 5
strict mode, end-to-end typing
Vue 3
for projects with an existing Vue codebase
also:
→
React + Next.js is the default for new products. Vue — when the client is already on Vue and migration isn't justified.
API
Backend
Async-first. Python and Go cover 95% of tasks. Node.js — for BFF or when the frontend team also writes backend.
primary tools
Go
high-throughput services, CLI, network utilities
Python 3.12
FastAPI, async, ML integrations
PostgreSQL 16
primary database, pgvector for embeddings
Node.js
BFF, real-time, unified JS stack
also:
→
Go for latency-critical services. Python for everything where development speed and ML/AI integrations matter.
MOB
Mobile
Cross-platform development as the default. Native — only when React Native objectively can't handle it.
primary tools
React Native
New Architecture, Bridgeless mode
Expo SDK 52
OTA updates, Expo Router, managed workflow
Swift
native iOS modules when JS isn't enough
Kotlin
native Android modules
also:
→
React Native covers 90% of mobile tasks. Native code — for Bluetooth, camera, heavy graphics.
OPS
Infrastructure
Infrastructure as code. Nothing lives only in one engineer's head. Every server, every pipeline — in the repository.
primary tools
Kubernetes
orchestration for products with 3+ services
Terraform
IaC, multi-cloud infrastructure
AWS / GCP
cloud providers
Docker
containerization, Docker Compose for dev/staging
also:
→
Docker Compose for small teams. K8s — when you have 3+ services and need autoscaling.
DAT
Data & AI
From ETL pipelines to production-grade LLM infrastructure. Data is the first architectural decision.
primary tools
pgvector
vector search on top of PostgreSQL
Elasticsearch
full-text and hybrid search
Языковые модели + поисковый контур
Claude, OpenAI, and open-source models
Kafka
event streaming, real-time pipelines
also:
→
pgvector — when you already have PostgreSQL and data < 10M. Qdrant — for heavy embedding workloads.
SEC
Security
Security isn't bolted on before production. Secure-by-default from the first commit. We design security into the architecture — not check boxes before release.
primary tools
also:
compliance readiness
SOC 2 Type II · PCI DSS · NBK requirements · AIFC compliance
security practices
Secrets rotation
automatic credential rotation every 90 days
Dependency scanning
CVE dependency checks on every PR
Container image scanning
Trivy in CI, blocking critical/high vulnerabilities
Audit logging
complete audit trail of all system actions
Zero-trust networking
service mesh, mTLS between services
Penetration testing
regular pentests with stakeholder reports
→
Security is not a pre-release checklist. Vault, OPA, Trivy are integrated into the process from sprint one.
Decision Framework
| Task | What we choose | Why |
|---|---|---|
| Landing / MVP | Next.js + Vercel | Fast start, managed infrastructure |
| Data-heavy dashboard | React + FastAPI + PostgreSQL | E2E typing, async I/O |
| High-load API | Go + K8s + Kafka | Latency, concurrency, scalability |
| AI product | Python + pgvector + языковая модель | ML ecosystem, native embeddings |
| Mobile app | React Native + Expo | Single codebase, OTA updates |
Have a project?
Tell us what you're building — we'll select the optimal stack and architecture.
hello@qdev.run →