allrider1127 Fastfetch Profile Banner

İTÜ OpenFOAM Python C++ Docker Linux Mint MIT

/*--------------------------------*- C++ -*----------------------------------*\
| ========= | |
| \\ / F ield | OpenFOAM: The Open Source CFD Toolbox |
| \\ / O peration | Version: v2512 |
| \\ / A nd | Website: www.openfoam.com |
| \\/ M anipulation | |
\*---------------------------------------------------------------------------*/
FoamFile
{
 version 2.0;
 format ascii;
 class dictionary;
 object allrider1127;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //
name "Jahongir Orzikulov";
affiliation "Istanbul Technical University - Mechanical Engineering";
roles ("Computational Mechanical Engineer" "CFD Analyst" "Systems Engineer");
focus "Compressible Flow · Solver Automation · HPC Pipelines · Propulsion";
stack
{
 simulation ("OpenFOAM" "CalculiX" "SU2" "Code_Saturne" "Elmer FEM" "NASA CEA");
 languages ("Python" "C++" "Bash" "C#" "C" "MATLAB");
 cad ("Onshape" "SolidWorks" "Fusion 360" "AutoCAD" "FreeCAD" "Blender" "Meshmixer" "FlashPrint" "LibreCAD");
 infra ("Docker" "Linux" "MPI" "Vim" "Conda" "Git");
 ai ("Ollama" "Hermes Agent" "Agentic AI" "SLM (Llama 3.2, Qwen)");
}
creed "Simulate it. Validate it. Automate it. Ship it.";
// ************************************************************************* //

PR Merged Lines Added Files Changed Tests Passed

/*---------------------------------------------------------------------------*\
| PR Details: |
\*---------------------------------------------------------------------------*/
mergeInfo
{
 repository "csml-rpi/Foam-Agent";
 status "Merged";
 linesAdded 1421;
 lab "NeurIPS 2025 · ML for Physical Sciences";
 paper "arXiv:2505.04997";
}
// ************************************************************************* //

Foam-Agent — a NeurIPS-published multi-agent framework for automating CFD simulations in OpenFOAM (arXiv:2505.04997).

• Engineered a 5-stage ESI translation middleware that enables Foundation v10 case files to run on ESI OpenFOAM containers (v2312, v2512) — eliminating the need for re-indexing the entire tutorial database
• Wired the translator into both the LangGraph pipeline and the MCP server, with post-translation state refresh to prevent stale graph nodes
• 10/10 pytest pass with regression coverage for Allrun bypass, regex scoping, and structural file remapping

"This is a very important contribution to the community and a very good design." — LeoYML, Foam-Agent maintainer (comment on PR #35)

OpenFOAM Mesh L/D Docker Turbulence

Full aerodynamic analysis of the Talon UCAV for SkyGuard Alliance — 1.7M cell hex-dominant mesh, converged at 1000 iterations. The entire simulation was orchestrated overnight using my KOR-Orchestrator daemon while I slept.

Why k-ω SST? — Industry standard for external aerodynamics. Accurately resolves near-wall boundary layers and adverse pressure gradients without wall functions, critical for predicting flow separation on swept delta wings.

Why simpleFoam? — Steady-state incompressible RANS solver. At 20 m/s freestream (Ma ≈ 0.06), compressibility effects are negligible — incompressible assumption holds.

Surface Pressure Distribution Velocity Field and Surface LIC

Left: Surface Pressure Distribution (p) — Right: Velocity Field & Surface LIC (U) @ 20 m/s

Python Zero Deps Linux Docker

A zero-dependency Python daemon that treats your personal workstation like a production compute node. Built entirely on Python's standard library — no pip packages, no YAML configs, no SLURM.

┌─────────────────────────────────────────────────────┐
│ 50°C 75°C 80°C 90°C │
│ ┃───────────┃───────────┃───────────┃ │
│ ▲さんかく ▲さんかく ▲さんかく ▲さんかく │
│ normal warning resume STOP │
│ │
│ Below 80°C → SIGCONT / docker unpause │
│ Above 90°C → SIGSTOP / docker pause │
└─────────────────────────────────────────────────────┘

• Thermal protection — auto-pauses workloads at 90°C via docker pause / SIGSTOP, resumes at 80°C
• Two-stage log scraper — detects FOAM FATAL ERROR in real-time across 10,000+ lines/sec of solver output
• Overnight mode — --mode shutdown with 30s Polkit-gated poweroff after last job completes
• Used this daemon to run the Talon UCAV simulation — queued the 1.7M-cell CFD job before bed, woke up to converged results

Ollama Llama Docker CPU Only

An autonomous OSINT pipeline that scrapes the SAHA 2026 defense expo exhibitor registry, cross-references company data, and uses a locally hosted SLM (Llama 3.2 3B) with native JSON enforcement to binary-classify 1,532 firms — filtering out finance/SaaS noise and scoring hardware engineering targets for internship relevance.

• Dual-process architecture — Playwright stealth scraper + LLM classification agent running in parallel
• Crash-safe & resumable — heartbeat checkpoints + SQLite WAL database, graceful kill switch
• CPU-only optimized — ~2GB RAM at inference on i7-12700H, keep_alive=0 for immediate model unload
• Result: 149 verified hardware engineering firms out of 1,532 exhibitors (9.7% signal-to-noise)

docker run --rm -it \
 --add-host=host.docker.internal:host-gateway \
 -v "$(pwd):/app" \
 allrider1127/saha2026-intelligence

24 hand-drawn engineering sheets produced for MAK 107 (Technical Drawing) at İTÜ — pencil on A3/A2 paper using T-squares, set squares, compasses, and mechanical pencils under strict ISO 128/129 standards.

⚡ Speed of Execution:

• Assignment 2 (Safety Valve Assembly — 10 detail drawings + 1 A2 assembly) — submitted 6 days after the assignment was announced. The rest of the class had the entire 13-week semester to complete it. First and fastest submission in the faculty.

Technical Text Drawing Safety Valve Assembly Drawing (A2)

Left: Assignment 1 — Technical Lettering on A3 (ISO) — Right: Assignment 2 — Safety Valve Assembly on A2 (ISO)

11 engineering sprints covering the full digital manufacturing pipeline: caliper metrology → parametric CAD → optical scanning → mesh repair → slicing optimization → rapid prototyping.

• Toolchain: OnShape, SolidWorks, Blender, Meshmixer, FlashPrint, AutoCAD, FreeCAD
• Hardware: EinScan SE-SP V2 3D Optical Scanner, FlashForge 3D Printer
• Physical-to-Digital: Every scanned part is backed by a version-controlled digital twin

Foam-Agent — csml-rpi/Foam-Agent • PR #35 • Merged

An end-to-end composable multi-agent framework for automating CFD simulations in OpenFOAM, published at NeurIPS 2025 Machine Learning and the Physical Sciences Workshop.

I implemented a runtime ESI translation middleware that post-processes Foundation v10 case files to run correctly on ESI OpenFOAM containers — a 5-stage deterministic mutation pipeline with JSON-based rule configuration, Allrun bypass filtering, and full pytest regression coverage.

/*---------------------------------------------------------------------------*\
| Publication Details: |
\*---------------------------------------------------------------------------*/
FoamAgentPaper
{
 title "Foam-Agent: Towards Automated Intelligent CFD Workflows";
 authors "(Yue Somasekharan Zhang Cao Chen Di Pan)";
 venue "NeurIPS 2025 ML for Physical Sciences Workshop";
 paper "arXiv:2505.04997";
 code "github.com/csml-rpi/Foam-Agent";
 myPR "#35 -> Merged (+1,421 lines, 14 files)";
}
// ************************************************************************* //

📎 Read the paper · 🔀 View the PR

• 🏛️ Istanbul Technical University — B.Sc. Mechanical Engineering (2024–2028)
• 🎓 YTB Türkiye Bursları — Full-ride scholarship + Üstün Başarı Burs Programı (double stipend)
• 🏫 Jizzakh Presidential School — Cambridge A-Levels (Physics, Chemistry, Mathematics), CIS Accredited
• 📊 SAT 1540 (Math 800/800) • IELTS 7.5
• 🇺🇿 Uzbek (native) · 🇬🇧 English (C1) · 🇹🇷 Turkish (C1) · 🇷🇺 Russian (elementary)
• 🛡️ SAHA 2026 Expo — UAE Pavilion Coordinator (ADNEC Group, EDGE, Calidus, TAWAZUN)
• ✈️ TEKNOFEST 2026 — SkyGuard Alliance (UCAV aerodynamics + Liquid rocket propulsion teams)

_{// built on linux mint · compiled with vim · shipped via git}