Go · C · Python · DevOps & Tooling
I build things at every layer — from bare-metal hypervisor sandboxes and Redis clones in C, to production SaaS backends and cloud automation pipelines.
📍 Cairo, Egypt · Open to Remote
I'm a backend and systems engineer with 3+ years of production experience.
I don't specialize in one layer. I've configured Xen hypervisors on bare metal, written TCP socket servers in C, designed multi-tenant PostgreSQL schemas with row-level security, and automated AWS EC2 build pipelines that replaced hours of manual work.
What drives me is understanding how things actually work — not just calling the API, but knowing what's happening underneath.
I'm currently deepening my embedded C and DevOps tooling skills, with a strong interest in systems-level roles in automotive and infrastructure engineering.
A from-scratch implementation of core Redis in C.
- RESP protocol parser
- TCP socket server
- Key-value store with TTL expiry
- Hash tables and dynamic arrays — no external libraries, manual memory management throughout
Graduation thesis — hypervisor-level malware analysis platform built on bare metal.
- Configured Xen hypervisor with 7 isolated Windows VMs (128GB RAM, NVMe, private bridged networking)
- Integrated DRAKVUF for agentless, kernel-level syscall tracing — invisible to running malware
- Built the feature extraction pipeline: syscall N-gram sequences, TF-IDF feature selection
- Dynamic XGBoost model achieved 93% TPR at 2% FPR on 71,000+ PE samples
My focus: infrastructure, sandbox orchestration, network isolation, and feature extraction pipeline
Production-grade multi-tenant SaaS for retail operations.
- PostgreSQL Row-Level Security with per-transaction
SET LOCALbusiness context isolation - Two-tier JWT system (generic + scoped tokens) with JTI blacklisting in Redis
- Fail-open Redis dependency — deliberate resilience decision
- 26+ REST endpoints: multi-branch inventory, POS sessions, role-based access control
Built for a digital publishing client — one-click desktop app generation.
- Launches on-demand Windows EC2 instances via UserData PowerShell scripts
- Full build sequence: pull template → inject book data → compile → upload .exe to S3 → update MongoDB → self-terminate
- 70% reduction in build costs — zero idle infrastructure, zero manual intervention
github.com/mu7ammad-3li/assist-plus
Distributed messaging API with AI integration.
- Multi-channel webhook processing (Facebook, Instagram) at scale
- Semantic caching layer via pgvector — 60% reduction in LLM API costs
- Async task system with Redis and BullMQ for reliable message delivery
github.com/mu7ammad-3li/cv-lize
NLP pipeline for resume analysis and candidate-job matching.
- spaCy NER for structured data extraction from unstructured resumes
- LLM-based scoring algorithm for candidate-job fit ranking
| Layer | Tools |
|---|---|
| Systems & Low-Level | C, Embedded C, sockets, memory management, RESP protocol |
| Backend | Go (Gin), Node.js (Express, Hono), Python (FastAPI) |
| Databases | PostgreSQL (RLS, pgvector), MongoDB, Redis |
| Cloud & Infra | AWS (EC2, S3, Spot Instances, IAM), Docker, Git |
| AI / LLM | Gemini, OpenRouter, spaCy, semantic caching, RAG, MCP |
| Frontend | React, Next.js, Electron, Flutter (AI-assisted) |
- Embedded C — microcontroller programming, memory-mapped I/O, bare-metal C
- DevOps tooling for embedded targets — build systems, cross-compilation, flashing pipelines
- Go deeper into systems programming
I'm the eldest in a family of eight. After losing my father, I took on the responsibility of keeping things running — while simultaneously building software, learning new stacks, and pursuing a technical career I believe in.
That context shapes how I work: I don't give up on hard problems, I'm honest about what I know and don't know, and I build things that actually work under pressure.
Open to backend, systems, DevOps, and embedded tooling roles.
