Return to the Oracle
DT
The Oracle's Keeper
Duygu Taş
Localization Project Manager · AI Workflows · Linguist
6 Years End-to-End
EN · TR · ZH · KO
Digital · Finance · Media · Games
LQA · MTPE · TM Governance
AI-Integrated QA · DTP
5,000+ LLM Outputs Evaluated
In the Oracle's Own Words
"I built this oracle because localization PM work is deeply invisible, and the craft deserves a mythology."
The frameworks, the judgment calls, the vendor rituals performed at midnight before a launch: they vanish the moment the project ships. Uncredited. Unseen. Nine cards to make them visible. This reading is for every PM who has held a multilingual release together with a briefing document and sheer will, and found no language for what that actually took. The cards found it instead.
"The machine speaks. I listen. I do not obey."
Human-first, AI-integrated. I treat artificial intelligence as a co-pilot that earns its seat through usefulness, not authority. Context stays human. Accountability stays human. Quality stays measurable. I design workflows where the machine handles volume and the practitioner handles judgment. The moment you stop questioning the output, you have stopped doing the work. Technology that explains itself and knows its limits. Applied at programme level.
The craft was always this ancient. Someone just finally gave it the ceremony it deserved.
The Craft: Skills & Disciplines
I
Translation Memory Management
TM leverage analysis, band discount negotiation, TM maintenance workflows
II
Linguistic Quality Assurance
MQM/LISA QA frameworks, in-country reviewer management, error typology
III
i18n & Pseudotranslation
Pre-translation testing, text expansion, RTL layout, string externalization
IV
MTPE & MT Strategy
Content tiering, DeepL/ModernMT integration, MTPE-L/F workflow design
V
Project Costing & Quoting
Word count analysis, fuzzy band modelling, budget forecasting, PO management
VI
Linguistic Asset Management
Termbase governance, style guide creation, TM ownership, brand voice
VII
Vendor Management
LSP briefing, RFP/RFQ, SLA definition, preferred vendor lists, KPI tracking
VIII
Localization Testing
In-context review, bug severity matrix, sprint integration, release criteria
IX
Transcreation & Cultural Adaptation
Transcreation briefs, back-translation review, cultural consulting, hreflang
Tools of the Trade
Phrase TMS
memoQ
SDL Trados
Crowdin
DeepL
ModernMT
JIRA
Confluence
Smartcat
XTM
Microsoft Excel
Figma
Currently Exploring
What the Oracle Is Reading
⟳
Agentic Workflows in Localization PM
Where does the PM's judgment end and the agent's automation begin? I am mapping the handoff points — scope confirmation, vendor selection, QA routing — where human oversight is not optional.
⌘
LLM Prompt Localization
A prompt is a text that needs to work across languages and cultures. Most people treat it as engineering. I think it is also localization — and the field has not caught up to that yet.
◈
AI Evaluation Frameworks for Linguistic Quality
After 5,000+ LLM output evaluations, the gap between automated metrics and actual linguistic quality is still the central unsolved problem. I am building frameworks that treat evaluation as a localization discipline, not a data science one.
◎
Turkish Market Localization Strategy
Agglutinative morphology, high uncertainty avoidance, a rapidly growing tech sector, and almost no dedicated localization strategy literature in Turkish. There is work to be done here — and I intend to do some of it.
✦
Interactive Localization Tools
This oracle is the first. The question I keep returning to: what else in the craft is invisible because nobody has found the right way to show it yet?
Open a Channel
Whether you are looking for a localization PM, want to discuss the craft, or simply made it this far and want to say hello: the door is open.