SO

INDEPENDENT BUILDER / NETHERLANDSSASCHA OROZ

01MAP THE PROBLEMFind the decision, dataset or workflow that actually limits the outcome.

02BUILD THE TOOLTurn that finding into software small enough to finish and operate.

03TEST THE RESULTCheck it against sources, numbers and real behaviour, not the demo.

  1. 01LokiOSBUILDING
  2. 02DatagateARCHIVED
  3. 03AI and Human-Capital RiskRESEARCH

INDEPENDENT BUILDER / NETHERLANDS

SASCHA OROZ

I build publishing software and research how AI changes work.

My background is in finance and strategy. I now build LokiOS, a private publishing system, and research whether firm-level AI exposure can partially hedge human-capital risk.

SEE SELECTED WORK

  1. 01MAP THE PROBLEMFind the decision, dataset or workflow that actually limits the outcome.
  2. 02BUILD THE TOOLTurn that finding into software small enough to finish and operate.
  3. 03TEST THE RESULTCheck it against sources, numbers and real behaviour, not the demo.

SELECTED WORK / 03

An active build, an archived venture and an open research question.

CORE WORKFLOW

  1. IDENTITY

    , supporting context: WRITER IDENTITY

  2. EVIDENCE

    , supporting context: SITE CONTEXT

  3. DRAFT
  4. REVIEW

    , supporting context: MONETIZATION POLICY

  5. VERIFY
  6. PUBLISH

    , supporting context: WORDPRESS

  7. MAINTAIN

    , supporting context: REPORTING

LokiOS workflow: seven stages in order, from identity to maintain, with five supporting context inputs.

012026-NOW

BUILDING

LokiOS

A private, self-hosted publishing system that takes an article from evidence to published page.

ROLE
PRODUCT, SYSTEM DESIGN AND IMPLEMENTATION
FIELD
AI workflows / Editorial systems / Product design / Publishing infrastructure

READ THE CASE STUDY

PUBLIC SIGNALS / UNSTRUCTURED

ANALYSIS

AI-SUPPORTED STRUCTURING

DECISION CONTEXT / STRUCTURED

  • SENTIMENT
  • NARRATIVES
  • SIGNALS
Datagate flow: many public signals converge into one AI-supported analysis layer, which structures three kinds of decision context. SCHEMATIC OF INTENT - NO PREDICTIVE ACCURACY IMPLIED

022023-2024

ARCHIVED

Datagate

An AI analytics MVP that organised noisy social-media conversation into research for digital-asset investors.

ROLE
CO-FOUNDER, STRATEGY AND PRODUCT
FIELD
AI analytics / Product strategy / Financial modelling / Go-to-market

READ THE CASE STUDY

EXPOSURE INPUTS

OCCUPATION × AI CAPABILITY

  1. 01OCCUPATIONAL AI EXPOSURE
  2. 02EMPLOYMENT COMPOSITION
  3. 03FIRM-LEVEL EXPOSURE
  4. 04PORTFOLIO CONSTRUCTION
  5. 05HEDGE EVALUATIONPARTIAL HEDGE TEST
Research design: an occupation-by-AI-capability exposure matrix feeds five stages in order. RESEARCH DESIGN - NO FINAL ESTIMATES OR PERFORMANCE SHOWN

032026-NOW

RESEARCH

AI and Human-Capital Risk

Research in progress on whether firm-level AI exposure can help build an equity portfolio that partially hedges AI-related human-capital risk.

ROLE
RESEARCH, DATA AND PORTFOLIO DESIGN
FIELD
AI exposure / Labor economics / Equity portfolios / Data integration

READ THE CASE STUDY

CURRENT SIGNAL / AUG 2026

  • 01BUILDINGLokiOSA private, self-hosted publishing system that takes an article from evidence to published page.
  • 02RESEARCHINGAI and human-capital riskWhether firm-level AI exposure can help construct an equity portfolio that partially hedges AI-related human-capital risk.
  • 03EXPLORINGLocal AI and small internet businessesWhat changes for one-person software businesses when capable models run on local hardware instead of the cloud.

I work on problems where a business question needs software to answer it: publishing workflows, portfolio construction, pricing.

  1. 01Define the constraintName the decision or bottleneck that actually limits the outcome before touching anything else.
  2. 02Show what is knownKeep measured facts separate from estimates, and mark what still waits on a decision.
  3. 03Verify where the work happensChecks belong inside the workflow, next to the person or agent doing the work.
  4. 04Price the model earlyHow a product earns money changes what it rewards, so monetization enters the design first.

CONTEXT / 2022-NOW

  1. 2022-2026

    Maastricht University

    INTERNATIONAL BUSINESS / FINANCE

    The financial grounding I still use: trade-offs, downside and uncertainty.

  2. 2022-2025

    Peakable

    STRATEGY & OPERATIONS

    Strategy and operations across more than 20 clients, mostly growth and financial modelling.

  3. 2023-2024

    Datagate

    CO-FOUNDER / STRATEGY & PRODUCT

    Co-founded an AI analytics venture: ten-week MVP, nine people, more than 20 investor pitches.

  4. 2025

    City University of Hong Kong

    EXCHANGE

    One semester of my International Business programme, studied from a different vantage point.

  5. 2026-NOW

    Independent work

    BUILDING / RESEARCH

    LokiOS, the human-capital research, and small experiments with local models.

WORKING NOTES

READ ALL NOTES

The short version: finance training, strategy work at Peakable, one archived venture, now building independently.

I am based in the Netherlands. I now build LokiOS, a private publishing system, and research whether firm-level AI exposure can partially hedge human-capital risk.

MORE ABOUT ME