Mission-critical platforms
Command centers, dispatch engines, realtime operational systems. The kind of thing that has to hold up under real load and real stakes, and keeps holding up.
Ibrahim Gamal · Solutions Architect & Engineering Lead
In the last year: a regional command center spanning 26 hospitals, delivered on deadline with a five-engineer team; an anti-hallucination RAG platform in production; the data-governance layer for regulated healthcare records. Most useful where the requirements are vague and the stakes are high.
Selected work
01
Jazan Health Cluster, government contract
NestJS · PostgreSQL · Prisma · Socket.io · Redis · LiveKit
The system of record fusing 26 hospitals into a single real-time operational picture: 71 live ambulances, an 8-state dispatch engine, and clinical KPI machines for STEMI, Stroke, and Trauma. Delivered on a government deadline with a team of five.
02
Multi-tenant patient flow, same cluster
Node.js · Express · Prisma · Next.js · SSE · Playwright
Live patient-flow aggregation across 26 emergency departments: a triage engine blending CTAS/ESI weighting with wait-time aging, race-safe daily token issuance, and public displays driven by server-sent events.
03
Autonomous LLM pipeline
n8n · OpenAI · Apify · Node.js
A 6-step autonomous pipeline that does the work of a senior content strategist: scrapes the niche, reasons over it with GPT-4, and ships production-ready calendars. Days of planning compressed to about 15 minutes per client.
04
Dual-sided legal marketplace
Next.js · Node.js · PostgreSQL · Stripe · Elasticsearch
A legal marketplace onboarding 500+ lawyers and facilitating 5K+ client matches: Stripe consultation payments, Elasticsearch legal discovery, real-time chat, and full Arabic RTL across client, lawyer, and admin tiers.
05
UN Climate Change, Bonn
Leaflet · PostGIS · PostgreSQL · Python · OpenStreetMap
A geospatial carpooling platform adopted organization-wide by the UN Climate Change sustainability team: PostGIS matching, Leaflet routing, and a deliberately zero-cost open stack so the UN could actually run it.
What I take on
Command centers, dispatch engines, realtime operational systems. The kind of thing that has to hold up under real load and real stakes, and keeps holding up.
Provider-agnostic agent layers, RAG grounded in your own data on PostgreSQL and pgvector, and cost-aware model routing. Production AI that answers truthfully and doesn't burn the budget.
Master-data governance for regulated records, GDPR control layers with full audit traceability, and legacy systems (CSV, SFTP, API) normalized into one model that holds 99.9% accuracy.
Seven sides of a person, mapped onto a 3D brain that physically re-forms into the meaning of each one. Ownership, the Adaptive Mind, the Builder, For Every Human, the Practice, the Mission, the Wanderer.
How an engagement runs
1
I sit in the business conversation, not just the ticket queue. Vague requirements are the starting point, not a blocker.
2
The system gets designed before it gets built: data model, integration backbone, failure modes, and the deadline math.
3
I write the hard parts myself and lead the team through the rest. The uncomfortable middle is where I am most useful.
4
Deployed, monitored, owned. If my name is on it, I own every layer, to the metal.
Contact
Vague requirements and high stakes are the brief, not the problem. Tell me what you are trying to make real.