Backend & Systems Engineer  ·  Founding Engineer @ Nover

"I build the unglamorous half of a product."

Backend & Systems Engineer. Final-year Electrical Engineering — Electronics & Communications student at Alexandria University, building production infrastructure now instead of waiting on graduation.

The APIs nobody sees. The queues that don't drop messages. The GPU pipelines that don't fall over at 3am. The schemas that survive the next migration. Production engineering and DevOps are where I'm strongest — AI infra, robotics, and embedded all lean on the same systems instincts.

Backend is where I go deep. I'm comfortable full-stack enough to ship a working UI over my own APIs — I just optimize for what's behind them.

Today — Founding Engineer & Tech Lead at Nover. I own the production backend for our AI image generation platform: fine-tuning diffusion models on H100s, serving inference through ComfyUI on AWS, and wiring it together with a custom WebSocket API that holds up under real traffic. Pre-seed in progress.

Recently — Software Engineer at RankUp (EU, remote). Built the topical map update system — an automated keyword research → clustering pipeline using SBERT embeddings and DataForSEO. Backend-and-data work, end to end.

Before — 1.5 years as a Software Engineer at a US neuromodulation research company. Built a real-time EEG digital twin pipeline: streaming brainwave data off OpenBCI hardware, running Hilbert transforms on the wire, and feeding it into coupled oscillator models. Authored a research paper on Kuramoto-based EEG synchronization along the way.

On the side — Led Mind Cloud (70+ engineers) from a 9th-place ERC finish to 2nd and 3rd at UGVC, and migrated the entire stack to ROS 2 — a distributed systems problem dressed up as a robotics one.

This is what I reach for first, and where most of my production work lives. Hover any icon for a tooltip.

Patterns I ship: REST · gRPC · WebSocket systems · long-lived streaming connections · idempotent task APIs.

Architecture focus: microservices · event-driven backends · queue-based workloads · atomic task distribution · real-time stream processing · ETL pipelines · structured logging.

What I actually do in production:

• Cloud architecture on AWS — ECS, S3, RDS, GPU instances for inference
• Container orchestration and reproducible Docker / Kubernetes pipelines
• Infrastructure-as-code with Terraform — repeatable, reviewable, rollback-friendly
• CI/CD with GitHub Actions — build, test, deploy, with rollback discipline
• Observability with Prometheus + Grafana — live dashboards, alerts, real traffic visibility
• GPU-backed service operation (H100s for diffusion) — cost, throughput, failure recovery
• Linux administration, shell tooling, and the boring glue that keeps systems alive

I lean on these when the system needs them — not the headline, but I've shipped real things with each.

Real repositories — distributed systems and production infrastructure first. Star counts update live.

Go Distributed Queue Reliable Task Queue High-performance distributed task queue in Go — atomic reliable queuing (RPOPLPUSH), concurrent workers, dead-letter queues, and a real-time dashboard. Dockerized for production.  View Repo →	Agent Mesh Distributed Task Orchestration Orchestration engine simulating an AI engineering squad — event-driven architecture over Go, Redis, and WebSockets, fronted by a zero-lag React command center (60 FPS @ 500+ tasks).  View Repo →
NetProbe Real-Time Traffic Analyzer Cross-platform C++20 packet analyzer — live capture (Npcap / libpcap), QUIC ClientHello SNI decryption, per-flow GeoIP/ASN, process attribution, and a Dear ImGui dashboard.  View Repo →	Real-Time Arbitrage Engine Streaming Data Pipeline High-frequency arbitrage detection with a decoupled microservices architecture — Apache Flink windowed analytics over Kafka streams, a Streamlit dashboard, and automated Discord alerts.  View Repo →
THAAD Simulation Physics-Based Missile Interception 3D interceptor sim with Kalman-filtered radar tracking, augmented proportional navigation, realistic drag / G-limits, and interactive 3D replays with battle reports.  View Repo →	Agentic LLM Architectures 12 Agentic Design Patterns Hands-on notebooks implementing reflection, ReAct, planning, multi-agent, memory, tree-of-thoughts & more — built with LangGraph, DeepSeek-V3, and Gemma 2.  View Repo →

CerebralFlow Neural Dynamics Framework "Digital twin" of brain dynamics — reconstruct functional networks from signals and run coupled-oscillator simulations with Hilbert transform analysis.  View Repo →	Physics-AI Bridge Statistical Mechanics ↔ AI Tracing the links between statistical mechanics and AI: a vectorized 2D Ising model mapped onto Hopfield networks to show how physical energy minimization drives memory recall.  View Repo →
Fourier Epicycle Machine Interactive DFT Visualizer Draw any continuous path and watch a machine of rotating epicycles reconstruct it in real time — the Discrete Fourier Transform, made tangible. Built with p5.js.  View Repo →	dot-spend Minimalist CLI Expense Tracker Cross-platform, minimalist expense tracker in pure Python — ASCII analytics, CSV export, and status-bar integration for Linux ricing.  View Repo →