work Selected Work

Real Systems, Running in Production

Twelve builds shipped end to end: multi-tenant SaaS, ERPs, secure record systems, agentic AI, fintech and native apps. Every one designed, built and deployed by our own team, the whole stack.

12Featured builds
7Domains
5Practice areas
14-Module ERP with a Native Field Sales AppERP & Systems
ERP & Systems2026

14-Module ERP with a Native Field Sales App

A complete ERP for an energy solutions business

business_centerEnergy solutions company

Stock, sales, purchasing, POS and double-entry accounting running in one system, handling roughly $300k of inventory movement a month. AI-driven reorder planning avoided about $20k of excess stock, and a native Android app put ordering in the hands of field salesmen.

  • 14 modules covering inventory, sales, purchasing, POS and accounting
  • AI reorder planning that reads movement history and flags what to order
  • Native Android app with offline access for field sales
  • Roughly $300k of inventory movement handled monthly
Next.jsFastAPIPostgreSQLAndroidDocker
~$20k excess stock avoided
Multi-Tenant RAG Chatbot PlatformChatbots & Support
Chatbots & Support2026

Multi-Tenant RAG Chatbot Platform

A full SaaS now used daily by 500 businesses

business_centerSaaS product

RAG-powered chatbots answering across web, WhatsApp, Instagram, Messenger and email, with a human-takeover inbox, lead capture and Stripe billing. Each tenant's bot answers from that tenant's own documents, with strict data separation between them.

  • Multi-tenant architecture with strict data separation
  • Five channels unified into one inbox
  • Human takeover mid-conversation with full history
  • Lead capture and Stripe subscription billing
Multi-tenant SaaSRAGVector searchWhatsApp APIStripe
500 businesses using it daily
Certification Company Business PlatformERP & Systems
ERP & Systems2026

Certification Company Business Platform

One platform running a GCC certification business end to end

business_centerGCC certification company

Certifications, exams, a sales team, finances and outbound in a single system. The rebuild took successful external attacks to zero, migrated more than 30,000 legacy records, and supported SAR 600,000 of pipeline in one month with SAR 250,000 closed.

  • Certification and exam management as the platform core
  • Sales pipeline connected to certification records
  • Finance module and outbound tooling for the sales team
  • Security rebuild closing the routes the previous system left open
Full-stack buildPostgreSQLSecurity hardeningSelf-hosted
Successful attacks reduced to zero
AI Recruitment Screening AgentAI & Automation
AI & Automation2026

AI Recruitment Screening Agent

CV scoring, then a live voice interview on the call

business_centerRecruitment agency

A CV is scored against the job description, and on a pass the agent joins the Google Meet, Zoom or Teams call itself, runs a live voice interview, and returns a structured scored report with the transcript attached.

  • CV scored against the job description with reasoning attached
  • Agent joins Meet, Zoom or Teams automatically on a pass
  • Live voice interview covering availability, notice and expectations
  • Structured scored report returned with full transcript
Voice agentLLMMeet / Zoom / TeamsPython
First-round screening automated
AI SEO Console with Git-Based PublishingSEO & Growth
SEO & Growth2026

AI SEO Console with Git-Based Publishing

Page one on almost every target keyword in under 30 days

business_centerSaaS company

An internal ops console combining an AI SEO and content assistant, a keyword tracker and a content Kanban, publishing straight to the live marketing site by committing MDX through GitHub. No shared database and no CMS.

  • AI SEO and content assistant for drafting and optimising
  • Keyword tracker showing movement against the target set
  • Content Kanban from draft through review to live
  • Publishing by committing MDX to GitHub, no CMS layer
Next.jsMDXGitHub APILLMSearch Console
Page one in under 30 days
Agency Operating SystemAI & Automation
AI & Automation2026

Agency Operating System

One platform replacing six separate subscriptions

business_centerMulti-service agency, 20 people

An AI-native operating system that replaced Slack, Zoom, cloud storage, team monitoring, client portals and ClickUp for a 20-person agency. One typed core serves web, mobile, desktop, a Chrome extension and an MCP server, with AI reading leads, team performance and task priority on top.

  • One typed core serving five surfaces from the same codebase
  • Client portals, tasks, storage and comms in a single system
  • AI layer reading leads, team performance and task priority
  • Six per-seat subscriptions retired
TypeScriptNext.jsMCPChrome extensionSelf-hosted
Six tools consolidated into one
Live Spoken-English Voice TutorChatbots & Support
Chatbots & Support2026

Live Spoken-English Voice Tutor

A real-time voice agent teaching a six-year-old to speak

business_centerEducation product

An animated character holds a spoken conversation with a young child, drops into her mother tongue when she is stuck, shows flashcards, awards stars, and remembers previous days.

  • Real-time voice conversation with an animated character
  • Falls back to mother tongue when the child stalls
  • Flashcards and star rewards for visual reinforcement
  • Session memory carrying across days
Speech-to-textLLMText-to-speechReal-time audio
Real-time bilingual voice
Agentic Swing-Trading SystemAI & Automation
AI & Automation2026

Agentic Swing-Trading System

Analyst agents, a bull-bear debate, and a hard risk engine

business_centerTrading systems R&D

Technical and news analyst agents plus a structured bull-versus-bear debate feed a trader agent, gated by a pure-Python risk engine that can only veto or shrink a position. Paper execution only, fully journaled.

  • Specialist analyst agents for technical indicators and news
  • Structured bull-versus-bear debate rather than averaged opinion
  • Deterministic Python risk engine that can only veto or reduce
  • Every decision journaled with its reasoning
PythonMulti-agentLLMDeterministic risk layer
Paper execution, fully journaled

Have something like this you want built?

Tell us what you are trying to ship. We build the whole stack, from the database to the AI to the deployment, and you work directly with the engineers writing the code.

Scroll to Top