I'm Jalalledin "Moji" Taavoni — freelance AI Integration Architect and Data Engineer based in Milano 🇮🇹.
I take AI from notebook demos to systems that run: reliably, observably, at the right cost, sometimes on the edge. The hard part isn't the model — it's the integration: stable orchestration, eval gates that fire before users notice, guardrails that hold, and inference that can run on a laptop when the cloud isn't an option.
const moji = {
role: ["AI Integration Architect", "Data Engineer", "DataOps Consultant"],
stack: ["Azure", "Databricks", "LangChain", "PromptFlow", "Foundry Local",
"SQL Server", "Synapse", "Fabric", "dbt", "Neo4j"],
philosophy: "Thoughtful before fancy.",
education: "Computer Science + Digital Humanities · Università di Pisa",
currently: "Building production LLM systems & agents on Azure",
open_to: "Freelance projects · IT and Remote EU",
reach: ["mojitmj.github.io", "linkedin.com/in/mojitmj", "t.me/mojitmj"],
};☁️ Cloud · Platform · DevOps
🧠 AI · ML · LLMs
💾 Data Engineering · DataOps
💻 Languages · Tools
|
PowerShell tool that x-rays a SQL Server / Azure SQL instance in one command — full DDL, DMVs, backup history, security audit, design-quality checks, per-table data samples. Cross-platform schedulers (Task Scheduler · SQL Agent · SSIS · cron · systemd).
|
GitHub Actions workflow for Azure Data Factory — JSON schema validation, hardcoded-secret scanning, Key Vault enforcement, trigger-state gates on every PR. Drop-in for any ADF estate.
|
|
Digital-humanities side project: 175 years of Italian academies as a property graph in Neo4j, visualized in the browser with popoto.js. Where data engineering meets the archive.
|
Live portfolio: dual-positioning landing page (AI / DataOps / DE / BI / DA), animated streaming-source boot, EN/IT toggle with Italian-flag theme, live chat overlay, full visitor metadata pipeline.
|
From: 23 May 2026 - To: 30 May 2026
Total Time: 51 hrs 49 mins
SQL 11 hrs 31 mins █████▒░░░░░░░░░░░░░░░░░░░ 21.03 %
JSON 8 hrs 54 mins ████░░░░░░░░░░░░░░░░░░░░░ 16.24 %
PowerShell 8 hrs 48 mins ████░░░░░░░░░░░░░░░░░░░░░ 16.07 %
Markdown 8 hrs 32 mins ████░░░░░░░░░░░░░░░░░░░░░ 15.58 %
JavaScript 3 hrs 44 mins █▓░░░░░░░░░░░░░░░░░░░░░░░ 06.81 %
Batchfile 2 hrs 17 mins █░░░░░░░░░░░░░░░░░░░░░░░░ 04.19 %
XML 1 hr 53 mins █░░░░░░░░░░░░░░░░░░░░░░░░ 03.45 %
- 🔒 Closed issue #1 in mojiTMJ/mojiTMJ
- [Day 21 - CI/CD Fundamentals](https://dev.to/17j/day-21-cicd-fundamentals-4aeg) Mon Jun 01 2026 4:21 AM- [I Let Hermes Agent Run My Workflow for a Week — Here's What Actually Happened](https://dev.to/prshant01/i-let-hermes-agent-run-my-workflow-for-a-week-heres-what-actually-happened-3hk5) Mon Jun 01 2026 4:18 AM- [We Ran 4 Claude Code Dialogs for 28 Hours. Here's What the Memory Layer Caught (and Missed).](https://dev.to/chunxiaoxx/we-ran-4-claude-code-dialogs-for-28-hours-heres-what-the-memory-layer-caught-and-missed-27p8) Mon Jun 01 2026 4:14 AM- [Understanding Jenkins CI/CD Using a Tiny Java Project (A Beginner-Friendly Walkthrough)](https://dev.to/sanjayghosh/understanding-jenkins-cicd-using-a-tiny-java-project-a-beginner-friendly-walkthrough-6co) Mon Jun 01 2026 4:13 AM- [Reflection SDD: Use a Reflection Harness to Level Up Your OpenSpec Workflow](https://dev.to/qtalen/reflection-sdd-use-a-reflection-harness-to-level-up-your-openspec-workflow-15l7) Mon Jun 01 2026 4:00 AM
- 🤖 Production AI — taking LLM demos / RAG / agent prototypes to systems that survive Tuesday morning
- 🛡️ AI evaluation & guardrails — golden sets, drift detection, regression gates, jailbreak hardening
- ⚡ Edge AI — Azure AI Foundry Local · ONNX · on-device LLMs for latency- or privacy-bound workloads
- 🏗️ DataOps / Data platform — lakehouse design on ADF + Databricks, CI/CD, governance, FinOps
- 🔧 SQL Server modernization — legacy → Azure SQL / MI / Fabric with replayable migrations
- 📊 BI / Power BI rescues — slow reports, wrong numbers, ungoverned sprawl
shipping: production AI integrations on Azure for SMB clients in IT/EU
building: sqlsnapshot v2 with Azure SQL DB + Fabric warehouse coverage
exploring: on-device LLMs (Phi-3, Llama-3) via Foundry Local + ONNX
reading: "Designing Data-Intensive Applications" (annual re-read)
learning: proper formal verification for AI agents
sipping: a long espresso ☕
"Pipelines should be like good plumbing — invisible, reliable, and someone else's problem when they break. I'm the someone else."
🟢🤍🔴 Fatto con espresso ☕ & 🍕 a Milano · Ciao mondo!