$whoami --context=micro-system

Anthony Lazkani

[software engineer]Software Engineer

I explore how complex systems operate at the micro level.

Software engineer building voice pipelines that answer in 0.88s, terminal-native AI agents shipped to npm, and full-stack systems where the details actually hold together.

statusonlinelocationGif-sur-Yvette, Francefocusmicro-systems
▶view_systeminspect_modules
01// module_registrytrace: scanning

The Module Registry

MOD_01core runtimes

Languages

MOD_02app shells

Frameworks

MOD_03intelligence layer

AI / Data

MOD_04infrastructure

Tools

02// deployed_modulestrace: idle

Deployed Modules

mailsh — npm @armjunior/mailsh

Mailsh

NEW

A globally published NPM CLI tool. Acts as a terminal-based AI assistant that triages inboxes, drafts emails via LLM, and manages contacts directly from the command line using local database resolution.

$npx @armjunior/mailsh --inbox

›resolving local index ......... done

›14 threads · 3 urgent · 2 from AI-ops

>draft 1: "Re: CI pipeline hotfix" — [y/n]

✓reply drafted · contact updated in sqlite

▊
experts — signal fusion

Financial Advisory Bot

Mixture-of-Experts LLM architecture. Intelligent signal fusion (news, market, alphas, fundamentals) for decision support, with performance evaluation driven by trading strategies.

$./experts run --fusion

›loading experts ████████░░ 4/6

›[news] bullish (+0.72)

›[market] neutral (+0.08)

›[alphas] bullish (+0.41)

→fused signal: LONG · conf 0.62

✓strategy backtest · Sharpe 1.84

▊
friendshare — trip.split

Friendshare

Full-stack expense-sharing app with trip management, custom split algorithms, and multilingual translation integration.

$friend split --algorithm fair

›trip "Bali · Aug" · €1,284.50

›anthony → €412.00 · salma → €297.50

›leo → €312.00 · maria → €263.00

✓settled · zero-debt engine

▊
03// system_pipeline

System Pipeline

  • NODE_01Jan 2025 — Jul 2025

    Software Engineering Intern

    Saint Joseph University·Beirut

    • ›Developed Android backend (FastAPI) for home automation with Voice Auth.
    • ›Integrated AI voice recognition (SpeechBrain) and an NLP model (Scikit-learn, 85.85% accuracy).
    • ›Parallelized processing layers, cutting inference latency to 0.88s.
  • NODE_02May 2024 — Jul 2024

    Information Technology Intern

    Dar Al-Handasah·Beirut

    • ›Configured MicroTik routers, VLANs, and ESXi Windows Server VMs.
    • ›Developed a frontend social network with React.js and real-time updates.