Skip to main content

LIOR · AI ENGINEER / DIGITAL TRANSFORMATION

I turn operating problems into intelligent systems that run.

From data engineering and operating analytics to local-first AI agents, I translate ambiguous requirements into systems that can be verified, maintained and used.

View selected work Decision records
Shanghai · Remote Data · AI · Systems Local-first by default

01 Positioning

Engineering intelligence, from problem definition to verification.

My work sits between operating context and engineering delivery. I align definitions, structure the data, choose the model and system boundaries, and keep a clear path back to evidence.

  • 01 Operating analytics and metric governance
  • 02 Data architecture and decision interfaces
  • 03 Local-first AI and agent workflows
  • 04 Engineering evaluation of AI tools

02 Selected Work

Systems built around real operating constraints.

Enterprise work is anonymized. Each case separates what was implemented from what is reconstructed for public presentation.

  1. Contract Profitability Management System project interface 01 / 04
    Business Intelligence 2025 In production

    Contract Profitability Management System

    A contract-level operating intelligence system that reconciles revenue, cost, collection and delivery data into one traceable profitability model.

    Role
    Led solution design and delivery across metric definitions, contract-level data modeling, warehouse layers, management dashboards and acceptance rules.
    Outcome
    Replaced disconnected monthly summaries with a traceable path from management-level indicators to individual contracts, projects and cost sources.
    View full case
  2. Digital Operations Analytics for Industrial Services project interface 02 / 04
    Digital Transformation 2025 Ongoing

    Digital Operations Analytics for Industrial Services

    An operating analytics foundation that moves industrial-service management from manually assembled monthly reports to consistent, queryable daily intelligence.

    Role
    Owned the end-to-end operating analytics program, translating between business definitions, data architecture and role-specific decision interfaces.
    Outcome
    Established shared metric definitions, a maintainable data pipeline and management views that expose operating issues earlier.
    View full case
  3. Local-first AI Agent Workspace project interface 03 / 04
    AI Engineering 2025 Ongoing

    Local-first AI Agent Workspace

    A privacy-aware AI workspace that combines local models, tool orchestration and agent workflows while keeping sensitive data on-device by default.

    Role
    Independently designed the model routing, tool orchestration, agent workflows and reliability controls for the complete workspace.
    Outcome
    Created a reusable local-first workflow that reduces cloud dependency without giving up access to stronger models when a task truly requires them.
    View full case
  4. Engineering Evaluation System for AI Tools project interface 04 / 04
    AI Engineering 2024 Ongoing

    Engineering Evaluation System for AI Tools

    A repeatable evaluation system for comparing AI tools and models against fixed tasks, criteria and evidence rather than demos or subjective impressions.

    Role
    Designed and maintained the evaluation framework, benchmark tasks, assessment dimensions and versioned comparison records.
    Outcome
    Turned “which tool feels better” into a traceable selection process that exposes failure boundaries and identifies the right tool for each workflow.
    View full case

03 Writing

Three lines of inquiry, developed through practice.

  1. 01

    Local-first Systems

    Data boundaries should be an explicit engineering choice.

    Local models, layered routing, privacy constraints, and the cases where local inference is the wrong trade-off.

    Read in Chinese
  2. 02

    Operating Intelligence

    Align the operating language before building the interface.

    Metric definitions, data governance, traceability, and the sequence that turns dashboards into decision systems.

    Read in Chinese
  3. 03

    AI Engineering

    Agents need recoverable workflows, not better demos.

    Repeatable evaluation, state, permissions, failure recovery, and the points where human judgment must remain.

    Read in Chinese
Portrait of LIOR
LIOR · Shanghai, China

04 About

I care more about judgment than tool accumulation.

I work across enterprise digital transformation and AI engineering. I start by defining the decision and its constraints, then build the data, model and system path needed to support it.

I prefer local-first, evidence-led and maintainable systems. The technology can change; the responsibility for boundaries, reliability and clarity remains.

CONTACT · SHANGHAI / REMOTE

Working on a problem worth treating seriously?

Send the current context, the decision or workflow that is not working, and the result you need. I will reply if my experience is a good fit.

Start a conversation