I build production AI systems and the engineering organizations around them.

I'm currently AI Practice Lead at Expleo Group, leading Data & AI work across enterprise clients: agentic workflows, RAG systems, LLM governance, human-in-the-loop automation, AI adoption, and distributed engineering teams.

My focus is simple: move AI from demos into real workflows that are useful, measurable, governed, and safe to operate.

• Production GenAI systems - RAG, agentic workflows, LLM orchestration, function calling, semantic search, and AI workflow automation.
• AI governance and readiness - RBAC, human-in-the-loop approvals, evals, observability, token and cost control, compliance gates, and operating models for AI adoption.
• Engineering leadership - distributed teams, async delivery, team leads, career frameworks, roadmap ownership, and platform engineering.
• Privacy-first AI architecture - local-first workflows, provider abstraction, anonymization, audit trails, and safe use of public or local model providers.
• Applied AI - AI systems that solve real operational problems in sales, legal, HR, engineering, and enterprise delivery.

Open-source CLI and documentation kit for assessing whether an AI workflow is ready for production.

It helps teams evaluate business value, data readiness, RAG quality, model architecture, governance, human review, evals, observability, cost, operations, and adoption.

Why it matters: most AI demos fail after the prototype because nobody has checked the operational parts.

Repo: https://github.com/mihaibc/ai-production-readiness-kit

Local-first AI privacy gateway, being rebuilt in Rust.

The goal is to detect and anonymize sensitive data before prompts are sent to local or cloud LLM providers, with provider-agnostic routing, policy controls, reversible tokenization, and audit reports.

Why it matters: companies should not have to choose between blocking AI adoption and leaking sensitive data into tools they do not control.

Repo: https://github.com/mihaibc/PrivacyCopilot

Automation scripts and utilities for AI/LLM workflows, file management, local models, prompt preparation, and developer productivity.

Repo: https://github.com/mihaibc/automate_everything

I'm currently building around three ideas:

1. AI production readiness
   How teams decide whether an AI workflow is still a demo, ready for pilot, or safe enough for production.
2. Privacy-first AI enablement
   How companies can safely use local and cloud models without exposing sensitive data.
3. AI-assisted engineering leadership
   How engineering teams use AI without losing architecture quality, ownership, testing, and maintainability.

• AI Compatibility Debt - why GenAI demos work but production AI systems break when teams ignore evals, governance, observability, model dependency, and ownership.
• AI Production Readiness - how to move AI workflows from prototype to governed production.
• AI-ready engineering organizations - how data, platform, product, and engineering teams need to change for AI to create real value.

LinkedIn: https://www.linkedin.com/in/mihaibc

AI / LLMs: Azure OpenAI, OpenAI API, Anthropic, Gemini, LangChain, Semantic Kernel, Google ADK, Ollama, pgvector

Backend / systems: .NET/C#, Python, TypeScript, Go, Rust, PostgreSQL, Docker, Kubernetes, Terraform

Cloud: Azure, Google Cloud, AWS

Leadership: engineering org design, distributed teams, AI adoption, roadmap ownership, C-level communication, technical presales

I like comparing notes with people working on production GenAI, AI governance, privacy-first AI, and engineering organizations in the AI era.

LinkedIn: https://www.linkedin.com/in/mihaibc