Skip to main content

About

About

I work between operating context and engineering delivery: defining the problem, structuring the data, choosing the boundary and leaving a system that can be maintained.

Portrait of LIOR
LIOR AI Engineering · Shanghai

Engineering intelligence with explicit boundaries.

My day-to-day work spans enterprise digital transformation, data engineering, contract profitability, operating analytics and data governance. In parallel, I build local-first AI and agent workflows for real work rather than isolated demonstrations.

The work usually starts with an ambiguous operating problem. I align the language and ownership, create the data and engineering structure, and deliver a system that can be used, reviewed and extended.

I care about maintainability, recoverability and evidence boundaries. Tools change quickly; responsibility for the system does not.

Role
AI Engineer · Digital Transformation
Based
Shanghai · Remote
Default
Local-first · Evidence-led

System quality is decided before implementation.

  1. 01

    Start with the decision

    Define the operating question, owner and evidence boundary before selecting tools.

  2. 02

    Keep the path to evidence

    A metric or model output should remain traceable to its source, definition and limits.

  3. 03

    Local-first by default

    Treat data export as an explicit choice, not the default cost of using an AI system.

  4. 04

    Design for maintenance

    Prefer recoverable workflows, replaceable tools and structures that survive the demo.

Where I can contribute.

01

Operating Intelligence

Contract-level profitability, governed metrics and decision interfaces.

02

Data & Digital Architecture

Data pipelines, warehouse layers, governance boundaries and implementation sequencing.

03

Local-first AI

On-device inference, model routing, agent workflows and privacy-aware tool orchestration.

04

Engineering Evaluation

Repeatable tasks, failure boundaries and evidence-led selection of AI tools.

CONTACT · SHANGHAI / REMOTE

Bring the operating problem, not a tool request.

Send the current context, the constraint that matters and the result you need. I will first determine whether my experience is a good fit.

Start a conversation hello@liorbase.com