Software Engineering

We Build Software That Ships

Backend systems, cloud infrastructure, AI pipelines, and enterprise interfaces — designed, built, and operated end-to-end by a senior engineering team.

0+
Years
0
Industries
0+
Production Systems
0
Continents

Why ForkTex

Own Infrastructure

We deploy on our own cloud platform with blue-green releases, encrypted vaults, and real-time monitoring — no third-party PaaS lock-in.

AI-Ready Delivery

Every project can tap into our AI pipeline tooling — from vector search and RAG to multi-modal LLM processing, built in-house.

Full-Stack Ownership

One team owns backend, frontend, infrastructure, and operations. No handoffs, no silos — from first commit to production monitoring.

Delivering Across

RomaniaNetherlandsUnited StatesCanadaDenmarkEuropeNorth AmericaRomaniaNetherlandsUnited StatesCanadaDenmarkEuropeNorth AmericaRomaniaNetherlandsUnited StatesCanadaDenmarkEuropeNorth America

Archetypes

Proven platform blueprints. Every delivery is assembled from these building blocks.

API

API

Async FastAPI services with SQLAlchemy, auto-generated OpenAPI schemas, JWT auth, structured errors, and Alembic-managed migrations. The backbone archetype.

FastAPISQLAlchemyOpenAPIAlembicJWT
CLI

Client

One codebase, three targets — iOS, Android, web — built on Expo + React Native with RTK Query bindings auto-generated from the API.

Expo 54React NativeRTK QueryCross-Platform
WEB

Web

Server-rendered Next.js apps with type-safe auto-generated API hooks, standardized state management, and schema-driven data fetching — zero hand-written fetch code.

Next.jsSSRAuto-Generated HooksType-Safe
DB

DB

PostgreSQL 16 with Alembic migrations and SQLAlchemy models. Multi-tenant aware, transactional, with per-service backups via systemd timers.

PostgreSQL 16Alembicpg_dumpMulti-Tenant
STO

Storage

S3-compatible object storage (MinIO local, S3 cloud) with multi-bucket access isolation, presigned URLs, and a clear dev-to-prod migration path.

MinIOS3-CompatiblePresigned URLsAccess Isolation
VEC

Vector

Qdrant-backed semantic search and embedding pipelines, with tenant-scoped retrieval, hybrid search, pluggable model providers, and content-aware chunking.

QdrantHybrid SearchTenant-ScopedPluggable Models
QUE

Queue

PostgreSQL-backed background jobs — no Celery, no Redis broker. One database is enough. Scheduled, retryable, observable through the same DB you already run.

PostgreSQL JobsNo CeleryRetryableScheduled
MAI

Mail

Self-hosted Stalwart mail server with PostgreSQL-backed directory, async aiosmtplib senders, Jinja2 templates, and SPF/DKIM/DMARC from day one.

StalwartaiosmtplibJinja2SPF/DKIM/DMARC

ForkTex Standards

Non-negotiable engineering standards applied to every system we deliver.

MA

Manifest + FSD

Every project ships a forktex.json manifest and declares an FSD maturity level (L1 Runnable → L4 Operational). 21 atoms across code, data, infra, ops — each mapped to one Make target.

QU

Quality Pipeline

Format, lint, typing, test, security, license, sync — all gated on every merge. No mocks: tests run against real PostgreSQL, real Redis, real services in containers.

AP

API Codegen

OpenAPI schemas drive type-safe RTK Query hooks for web and client. Two-layer architecture — generated bindings + hand-written orchestration. Zero hand-written fetch code.

HO

Hot Reload

Containerized dev with mounted sources for instant feedback. Code changes reflect immediately, no rebuilds. The full stack starts with one command.

WO

Work Analytics

Hourly-grained time tracking per project and member, surfaced inside ForkTex NetWork. Weekly through yearly breakdowns with full activity transparency.

MA

Makefile

One Makefile per project, one Make target per FSD atom. `make install`, `make test`, `make apply` — the same verbs across every repo we ship.

DO

Documentation

README, architecture, runbooks, ADRs — committed alongside code. Generated manuals (C4 diagrams + filesystem inspector + AI-consumable context bundle) on `make manual`.

AG

Agent Recall

Codebase knowledge is captured as structured concepts, rules, and few-shots — consumable by humans and downstream AI agents for consistent project context.