Data Management Manager · Legal Reporting · Istanbul, Turkey
Onur
Altıntaşlı
Data Management Manager with 13+ years of experience in Turkey's finance, insurance, and pension sectors. Currently leading legal reporting operations and data governance at Anadolu Hayat Emeklilik — Turkey's largest life insurance and pension company — with end-to-end ownership of statutory compliance pipelines including FATCA, CRS, HAYMER, and Merkez Bankası reporting.
About
Data Engineer &
Governance Leader
Data Management Manager with 13+ years of experience in Turkey's finance, insurance, and pension sectors. Currently leading legal reporting operations and data governance at Anadolu Hayat Emeklilik — Turkey's largest life insurance and pension company — with end-to-end ownership of statutory compliance pipelines including FATCA, CRS, HAYMER, and Merkez Bankası reporting.
Combines deep technical expertise in Oracle PL/SQL, Java, and enterprise integration with hands-on team leadership and strategic data management. MSc in Engineering & Industrial Management from Marmara University..
13+
Years Experience
1M+
App Users
2
Sectors
7+
Certifications
Career
Work
Experience
Assistant Manager, Data Management
Anadolu Hayat Emeklilik · Istanbul, Turkey
- —Own end-to-end legal reporting pipelines for Turkey's largest life insurance and pension company — delivering full regulatory compliance across GEV, HAYMER, Merkez Bankası, FATCA, and CRS obligations with zero tolerance for error or delay.
- —Drive data management strategy and governance across the organization: data quality controls, lifecycle policies, and cross-department data operations.
- —Manage government contribution (devlet katkısı) processes and coordinate internal operational controls to ensure accuracy and auditability of all statutory submissions.
- —Lead and mentor a specialist team, bridging technical execution and stakeholder communication at the management level.
Senior IT Specialist & Software Architect
Anadolu Hayat Emeklilik · Istanbul, Turkey
- —Architected and delivered the GEV legal reporting system — a mission-critical Java application forming the statutory data pipeline between Anadolu Hayat Emeklilik and the Pension Monitoring Center (EGM).
- —Built and maintained enterprise-scale PL/SQL packages and procedures for large-volume financial data processing, ETL pipelines, and BI reporting.
- —Designed and deployed a Microsoft BizTalk Server integration layer to centralize payment procedures across multiple enterprise platforms — eliminating data silos and reducing manual reconciliation.
- —Delivered the FENIKS Oracle migration from AS400: complex ETL development handling mission-critical financial records with zero data loss.
- —Built end-to-end customer churn analytics pipeline and data lake infrastructure using Python and Knime, directly supporting strategic business decisions.
- —Developed Windows Forms, Windows Service, WCF, and C# applications for financial operations, billing, and internal tooling.
System Specialist
Emeklilik Gözetim Merkezi · Istanbul, Turkey
- —Core Oracle DBA and PL/SQL developer for Turkey's national Pension Monitoring Center — maintaining the database infrastructure underpinning pension data for millions of participants.
- —Delivered Oracle Forms applications, data mining solutions, and ERP management tooling for regulatory and operational use.
Solution Developer
Mobinex · Istanbul, Turkey
- —Built iOS and Android applications for Turkey's major financial institutions using the Smartface Financial Services Platform.
- —Developed Akbank Direkt — a full-featured mobile banking app that reached 1M+ active users — delivering the complete suite of banking operations on mobile.
- —Delivered mobile application for Sahibinden.com.
System Developer
Teknosa · Istanbul, Turkey
- —Oracle ERP, CRM, and Siebel system development for one of Turkey's leading electronics retail chains.
- —PL/SQL form design, data mining, and Oracle Discoverer reporting.
Selected Work
Notable
Projects
01
GEV Legal Reporting System
Mission-critical Java application — the statutory pension data pipeline between Anadolu Hayat Emeklilik and Turkey's Pension Monitoring Center (EGM). Ensures full regulatory compliance for millions of pension records with zero tolerance for error.
02
Customer Churn Analysis & Data Lake
End-to-end analytics pipeline for customer churn prediction at Anadolu Hayat Emeklilik. Designed and built the data lake infrastructure from the ground up, enabling data-driven strategic decisions across the organization.
03
FENIKS Data Migration
Enterprise-scale Oracle database migration from AS400. Architected and executed comprehensive ETL scripts handling mission-critical financial records — delivered with zero data loss.
04
BizTalk Payment Centralisation
Microsoft BizTalk Server integration project using Oracle PL/SQL to centralize payment procedures across multiple enterprise platforms — eliminating data silos and unifying financial operations.
05
Akbank Direkt
Full-featured iOS/Android mobile banking app for Akbank, one of Turkey's leading banks. Reached over 1 million active users delivering a complete suite of banking operations.
Expertise
Skills &
Expertise
Data Management & Governance
Database & Backend
Development
Tools & Platforms
Leadership & Process
Credentials
Certifications
Background
Education
MSc Engineering & Industrial Management
Marmara University
2021 — 2023
Graduate-level focus on engineering management, operations, and organizational leadership
BSc Computer Engineering
Eastern Mediterranean University
Honor Certificate · Software Engineering Certification
Writing
Latest
Articles
2026-07-28
The Actuarial Assumption Change Problem: Why Updating a Single Mortality Table Rewrites the History of Every Policy That Depends on It
An assumption change in a life or pension book is not a forward-looking parameter update — it is a retroactive restatement event. Most pipelines were never designed to handle that, and the delta usually hides until the regulator asks.
Read →2026-07-27
The Consent Withdrawal Problem: Why KVKK Data Deletion Requests Break BES Pipelines That Were Never Designed to Forget
BES pipelines were architected to preserve every participant record for decades, but KVKK grants the right to erasure at any moment. Most firms handle this contradiction by pretending it doesn't exist.
Read →2026-07-26
The Participant Communication Data Problem: Why BES Disclosure Obligations Create a Reporting Layer Your Pipeline Was Never Designed to Prove
BES disclosure infrastructure in Turkey is almost universally built to send, not to prove. When a participant disputes what they were told three years ago, the gap between dispatch and evidence becomes a regulatory problem no one budgeted for.
Read →2026-07-25
The Regulatory Calendar Problem: Why BES Pipelines Break When Turkey's Holiday Schedule Shifts Late
Turkish public holidays get announced or amended with lead times that pension pipelines cannot absorb, and the resulting failure is not a missed deadline but a legally wrong participant balance. The fix is treating the holiday calendar as a temporal dependency in the event model, not as configuration.
Read →2026-07-24
The Segment Code Problem: Why BES Product Classification Is the Most Quietly Broken Layer in Pension Reporting
After a decade of owning the segment classification layer in a BES pipeline, I can tell you the real bug isn't in the mapping tables. It's that we've been modeling a temporally scoped legal designation as if it were a static dimension.
Read →2026-07-23
The Subrogation Data Problem: Why Insurance Recovery Rights Create a Parallel Ledger Your Pension Pipeline Was Never Built to Track
Subrogation recoveries retroactively change the net cost of claims your pipeline already closed months ago. That contradiction — settled claims that aren't actually settled — is where most insurance data architectures quietly break.
Read →2026-07-22
The Signature Authority Problem: Why BES Power of Attorney Records Break Pension Pipelines at the Worst Possible Moment
Power of attorney in BES systems is usually stored as a flag that gets overwritten, never versioned, and never tied to the legal event that created it. The failure surfaces at withdrawals, transfers, and death claims — exactly when the participant's family cannot absorb another delay.
Read →2026-07-21
The Regulatory Correspondence Problem: Why the Data Behind a Single Regulator Inquiry Takes Longer to Reconstruct Than the Answer It Proves
When EGM or BDDK asks about a submission from eighteen months ago, the answer takes minutes but the evidence takes weeks. Most firms built pipelines that produce outputs beautifully and preserve almost nothing about how those outputs came to exist.
Read →2026-07-20
The Intermediary Commission Problem: Why BES Agent and Distribution Channel Data Is the Most Underreconciled Layer in Pension Reporting
Commission attribution in BES looks like a finance workflow but behaves like a data lineage problem, and most pipelines were never built to handle it. After a decade of watching agent assignments drift silently across policy lifecycles, the reconciliation failures are structural — not accidental.
Read →2026-07-19
The Participant Age Verification Problem: Why BES Eligibility Calculations Compound Silently Until Retirement
BES eligibility is not a value you compute on demand — it is a legal state that evolves with every data correction upstream. Treating it as derived rather than recorded is why participants arrive at the counter expecting a retirement that your pipeline no longer agrees they earned.
Read →2026-07-18
The Batch Replay Problem: Why Reprocessing Historical Pension Data After a Logic Fix Is More Dangerous Than the Original Bug
Fixing a calculation error in a BES pipeline feels like a technical problem, but replaying the batch turns it into a legal one. The corrected output collides with state contribution records, EGM submissions, and NAV allocations that already settled — and the fix breaks more than the bug did.
Read →2026-07-17
The Regulatory Reporting Freeze Problem: Why BES Pipelines Cannot Process Normal Operations During Submission Windows
Submission windows are not a reporting problem, they are a concurrency problem. The pipeline is legally committed to a snapshot that must not change while production keeps writing to the same tables.
Read →2026-07-16
The Pension System Clock Problem: Why BES Pipelines Must Track Four Different Definitions of 'Now' Simultaneously
BES pipelines fail reconciliation not because of logic errors but because they collapse four independent temporal axes into one. After running HAYMER, GEV, and state contribution flows in parallel, the pattern is unmistakable: most pension data bugs are clock errors.
Read →2026-07-15
The Tax Withholding Timing Problem: Why BES Withdrawal Taxation Is Determined by Data Your Pipeline Recorded Years Ago
BES withdrawal taxation in Turkey depends on contribution-level data captured across years, most of which pipelines quietly overwrote or never versioned. When the withdrawal request lands, the tax engine often runs on approximations that nobody officially admits are approximations.
Read →2026-07-14
The Reference Data Decay Problem: Why Fund and Product Master Data Is the Silent Killer of Pension Reporting Accuracy
Fund codes and product parameters look static until a regulatory change proves they were never versioned properly. By then, years of HAYMER, GEV, and EGM submissions have already inherited the error.
Read →2026-07-13
The Parallel Submission Problem: Why Running FATCA, CRS, HAYMER, and GEV in the Same Calendar Window Is an Architecture Problem, Not a Scheduling One
Four statutory pipelines sharing a submission window is not a resourcing issue you solve with more infrastructure or a bigger team. It is a dependency graph problem, and treating it as a calendar problem is how correction cycles turn into contamination events.
Read →2026-07-12
The Restatement Problem: Why Correcting a Historical Regulatory Submission Is Harder Than the Original Filing
Fixing an already-accepted regulatory submission is not a technical patch — it is a legal event that ripples through every downstream reconciliation, ledger position, and dependent filing. Most firms only discover their pipelines were never built for restatement at the exact moment they need one.
Read →2026-07-11
The Inactive Participant Problem: Why BES Accounts That Stop Receiving Contributions Are the Hardest Records to Keep Legally Current
Inactive BES accounts don't leave the system — they drift silently while entitlements, valuations, and reporting obligations keep accruing. The quietest records in your database are usually the ones hiding a decade of unreconciled state.
Read →2026-07-10
The Death Certificate Problem: Why a Participant's Death Is the Most Legally Fragile Moment in BES Data
A death notification does not close a BES account — it opens a concurrent legal reconstruction that most pension pipelines are architecturally incapable of handling. The systems that ran flawlessly for twenty years of accumulation are usually the ones that fail silently the moment the death certificate arrives.
Read →2026-07-09
The Cross-Border Transfer Problem: Why a BES Participant Who Moves Abroad Breaks Every Assumption in Your Reporting Stack
A single BES participant relocating abroad triggers three regulatory regimes that were never designed to coexist inside one data pipeline. Most firms only discover the conflict when the CRS extract fails validation — years after the address change that caused it.
Read →2026-07-08
The Vesting Cliff Problem: Why BES Employer Matching Rules Create a Reconciliation Debt That Compounds Until Termination
Employer matching in corporate BES looks like a parallel contribution stream, but it is actually a deferred liability whose true state lives in HR, not in the pension system. By the time a termination triggers the vesting calculation, years of small misalignments have compounded into a legal dispute.
Read →2026-07-07
The Name Change Problem: Why a Participant's Legal Identity Update Is a Compliance Event Disguised as a Data Fix
A name change in a pension system is not a master data update. It's a retroactive identity resolution problem that touches EGM records, FATCA/CRS files, KVKK consent chains, and beneficiary designations signed under a name that no longer exists.
Read →2026-07-06
The Contribution Ceiling Problem: Why Annual BES Limits Create a Predictable Reconciliation Crisis Every December
BES contribution ceilings look like a trivial threshold rule until you meet the payroll data that feeds them. The real problem is architectural: most systems validate contributions one paycheck at a time when the regulation is fundamentally cumulative.
Read →2026-07-05
The Matching Problem: Why EGM Participant Identity Is Harder to Resolve Than Any Other Financial ID in Turkey
Bireysel emeklilik participant identity is not a lookup — it is a probabilistic match against decades of dirty master data, employer submission errors, and transfer records with conflicting truth. Most pension pipelines treat it as a join when it should be a confidence score with a legal audit trail.
Read →2026-07-04
The Employer Integration Problem: Why Corporate BES Contributions Break More Pipelines Than Participant Behavior Ever Does
Individual BES pipelines were designed around a single participant making decisions on their own timeline. Corporate contributions violate almost every assumption in that model, and the reconciliation failures are structural, not operational.
Read →2026-07-01
The Re-enrollment Problem: Why Automatic Enrollment Campaigns Are the Hardest Batch Operation in BES
Auto-enrollment looks like a marketing push from the outside. Inside the plumbing, it's a mass policy inception event that breaks every downstream system that was never designed to absorb it.
Read →2026-06-26
The Incomplete Withdrawal Problem: Why Partial Surrenders Break Pension Pipelines
Full terminations are painful but finite — they crystallize a policy's history into one binding calculation. Partial withdrawals leave the policy alive and retroactively rewrite the contribution history every downstream system already reported as settled.
Read →2026-06-25
The Fund Switch Problem: Why Participant-Directed Investment Changes Are the Most Underestimated Operational Risk in BES
A fund switch looks trivial on the screen but becomes a multi-party, time-boxed operation in production where mistimed cutoffs and reconciliation windows produce real-money discrepancies. The risk lives in sequencing assumptions made at design time and never revisited.
Read →2026-06-24
The Transfer-In Problem: Why Bringing a Participant's History from Another Pension Company Is the Hardest Data Operation in BES
A transfer-in is not a data import — it is the acceptance of another company's legal claims about a participant's history before EGM can confirm any of it. Everything that goes wrong downstream traces back to what the incoming file silently assumed about your schema.
Read →2026-06-23
The Surrender Value Calculation Problem: Why Policy Termination Is the Most Data-Intensive Moment in Life Insurance
Surrender is where every shortcut, retroactive adjustment, and quietly buried assumption in a policy's data history surfaces as a single legally binding number. The systems that handle inception and monthly contributions gracefully are often the same ones that collapse under termination.
Read →2026-06-22
The Dormant Account Problem: When a Pension Participant Disappears and the Data Has to Find Them
Turkish pension regulations assume you can always reach your participants, but the data infrastructure most companies built assumes the same thing. When the participant disappears, the pipeline doesn't bend — it breaks across compliance, data quality, and legal liability simultaneously.
Read →2026-06-21
The Beneficiary Designation Problem: Why Life Insurance Data Is Harder to Keep Correct Than Any Other Record
Beneficiary records look like a trivial lookup until a claim arrives and a decade of informal updates, missing signatures, and silent schema migrations become a legal dispute. The structural fix is to stop treating beneficiary state as a field and start treating it as an event log with legal standing.
Read →2026-06-20
The Retroactive Regulatory Change Problem: When the Regulator Rewrites the Rules on Data You Already Submitted
Regulators in Turkish pension and insurance don't just change rules going forward — they redefine how historical data must be interpreted. Most data pipelines are built to be correct at submission time, not correct as currently defined.
Read →2026-06-19
The Policy Anniversary Run: Why Batch Scheduling in Life Insurance Is Harder Than It Looks
Every active life and pension policy triggers a legally significant event on its anniversary, and scheduling those runs across millions of contracts is an architecture problem, not an ops one. After years of owning that pipeline, I can tell you the failures almost never live in the actuarial code.
Read →2026-06-18
The Reconciliation Gap Between Fund Valuation and Participant Balances: Why NAV Calculation Day Is the Most Dangerous Moment in Pension Data
NAV calculation gets all the audit attention, but the silent failures happen one layer down — in the translation between fund value and participant balance. After years of running that pipeline, I can tell you the risk lives where nobody looks.
Read →2026-06-17
The Cutover Window: Why Pension and Insurance System Deployments Cannot Follow Standard Release Engineering
Standard release engineering assumes the world holds still during your deployment window. In pension and insurance, contributions keep arriving, policies keep aging, and regulators do not pause their reconciliation cycles to accommodate your CI/CD pipeline.
Read →2026-06-16
The Contribution History Reconstruction Problem: When a Participant Disputes 10 Years of Records
A pension dispute over a decade of contributions is not a customer service ticket — it's a forensic reconstruction across every schema migration and retroactive correction your pipeline has ever made. Here is what it actually takes to produce a legally defensible point-in-time view.
Read →2026-06-15
The Participant Record as a Legal Document: Why Pension Data Cannot Treat Corrections the Way Other Systems Do
In Turkish individual pension, a correction is not a fix — it is a new legal event with retroactive consequences. Treating it like an UPDATE statement is how you end up with reconciliation breaks that take months to unwind.
Read →2026-06-14
The Schema Versioning Problem in Long-Lived Financial Contracts
A pension policy written in 2005 still pays out in 2026, but the data model that defined it has been overwritten a dozen times. Schema versioning in life and pension is not a DevOps hygiene issue — it is a legal liability that surfaces during audits, restatements, and cross-border reporting.
Read →2026-06-13
The Audit Trail Is Not a Log: Why Financial Regulators and Data Engineers Mean Different Things by the Same Word
When compliance asks for an audit trail and engineering points at the logging stack, both sides nod and walk away thinking the problem is solved. It isn't, and the gap shows up during the regulator's third question.
Read →2026-06-12
The Rollback Problem: Why Pension and Insurance Pipelines Cannot Afford the Recovery Strategies Everyone Else Uses
Standard rollback doctrine assumes a prior state can be restored and replayed forward. In pension and insurance pipelines, that prior state has already left the building — and trying to recover it the conventional way creates compliance violations.
Read →2026-06-11
When Stored Procedures Are the Right Answer and When to Get Out of the Database
The pendulum has swung too far toward application-layer logic, and regulatory reporting pipelines are paying the bill in latency and audit gaps. Knowing where logic belongs has less to do with architectural fashion and more to do with who gets the call at 2am.
Read →2026-06-09
KVKK Enforcement in Financial Data: What Is Genuinely Different from GDPR in Practice
KVKK is routinely treated as a Turkish translation of GDPR, but the enforcement mechanics and data residency realities in financial services tell a different story. Teams that build pipelines on that assumption end up rewriting them after the first KVKK Board decision lands.
Read →2026-06-08
Managing a Team That Cannot Be Wrong: Leadership Under Zero-Error Constraints in Regulated Finance
Leading a data team in pension and insurance is not about maximizing output or coaching performance. It is about engineering an environment where a single analyst's mistake physically cannot reach the regulator.
Read →2026-06-06
Data Quality Metrics That Actually Matter in Pension Data vs. the Ones Consultants Sell You
Generic DQ frameworks score completeness and accuracy on pension data and call it a day, but the real failures hide in effective dates, state transitions, and contribution bucket alignment. After running HAYMER, GEV, and state contribution pipelines side by side, the metrics worth operationalizing have almost nothing to do with what vendor dashboards display.
Read →2026-06-05
EIOPA Convergence in Practice: What Turkish Pension Companies Are Actually Building
EIOPA alignment is framed as a compliance project in Turkish insurance, but the real work is rebuilding data infrastructure that was never designed for the granularity and frequency these frameworks demand. The gap between SEDDK reporting habits and EIOPA-grade pipelines is where the actual engineering happens.
Read →2026-06-04
Payment Centralization Across Enterprise Platforms: The Hidden Complexity of Doing It Right
Centralizing payments looks like a clean consolidation win until you actually own the routing matrix, timing dependencies, and reconciliation surface that come with it. The decentralized mess you replaced starts to look like the easy problem.
Read →2026-06-02
Why Insurance Data Is Structurally Messier Than Banking Data — and How to Handle It
Banking data is transactional and linear; insurance data carries embedded uncertainty, multi-period state changes, and product structures that quietly break standard data quality frameworks. Here is what 13 years of life, pension, and health pipelines taught me about handling it.
Read →2026-05-31
Oracle DBA Skills Every Data Manager Should Have But Rarely Does
A data manager who cannot read an execution plan or interpret wait events is permanently dependent on DBAs they cannot evaluate. That dependency shows up in audit delays, broken pipelines, and architectural decisions made by the wrong people.
Read →2026-05-30
Zero-Error Tolerance in Statutory Submissions: How to Actually Achieve It Operationally
Zero-error tolerance is not a QA checkbox — it is a design constraint that dictates how you architect every pipeline upstream. Here is what actually works after running GEV, HAYMER, FATCA, CRS, and Merkez Bankası submissions in parallel.
Read →2026-05-29
The 9-Year Company Journey: What Staying Teaches You That Job-Hopping Cannot
The tech industry rewards movement, but regulated finance rewards memory. After 13 years at one insurance company, I've learned that depth compounds in ways breadth never will.
Read →2026-05-28
What Changes When You Move from Developer to Data Manager — and What Doesn't
The jump from senior developer to data manager gets sold as a soft skills problem. In regulated finance it's actually a trust architecture problem, and that's where most new managers quietly fail their first year.
Read →2026-05-27
PL/SQL in 2026: Why the Industry Keeps Declaring It Dead and Why It Keeps Being Wrong
Every few years a new generation of engineers declares PL/SQL obsolete, and every few years the statutory reporting pipelines that cannot fail keep running on it. That is not inertia. That is architecture.
Read →2026-05-26
Oracle Performance Tuning for Large-Volume Financial Data: What the Textbooks Skip
Standard Oracle tuning advice was written for OLTP workloads, not for pension valuation runs that scan 200 million policy rows under a six-hour SLA. Here's what actually works when batch windows, bitemporal joins, and regulatory reconciliations meet production.
Read →2026-05-25
What Regulators Actually Look for in Statutory Submissions — From Someone Who Sat on Both Sides
Compliance teams optimize for the wrong things in statutory reporting. After years of reviewing submissions at EGM and a decade preparing them, the patterns that trigger deeper scrutiny are rarely the ones companies prepare for.
Read →2026-05-24
HAYMER: Why Pension Data Pipelines Are Structurally Different from Every Other ETL You've Built
HAYMER submissions don't tolerate the assumptions baked into standard financial ETL. A misaligned timestamp or a missed state transition isn't a data quality ticket — it's someone's retirement balance.
Read →2026-05-21
Government Contribution Reconciliation: The Daily Complexity Nobody Writes About
The Turkish pension system's state contribution looks like a simple 30% match on paper. In production, it's a daily reconciliation problem where every timing mismatch turns into a real money discrepancy with EGM watching.
Read →2026-05-20
FENIKS: What Zero Data Loss Actually Means When You Migrate at Scale
Zero data loss is not a metric you celebrate at cutover. It is a constraint you bake into every schema decision, every reconciliation hook, and every rollback path from day one.
Read →2026-05-18
What BizTalk Taught Me That Modern iPaaS Tools Forgot
BizTalk is mocked as legacy bloat, but it forced architectural habits that today's iPaaS platforms quietly abandoned. The data quality failures we keep firefighting are the direct consequence.
Read →2026-05-17
FATCA/CRS Compliance: What the Data Pipeline Actually Looks Like
Compliance treats FATCA and CRS as a reporting obligation. The data engineers building the pipelines know it is a structural data quality problem dressed in regulatory language.
Read →2026-05-16
What 'Data-Driven' Actually Requires to Be True
Every organization claims to be data-driven. Very few have done the work that 'data-driven' actually requires. Here's the gap between the aspiration and the infrastructure.
Read →2026-05-14
The Governance Conversation You Should Have Before the Audit
Audits don't create data governance problems. They reveal them. The organizations that handle audits well had the governance conversation years before the auditors arrived.
Read →2026-05-12
Kafka in Finance: When Real-Time Actually Matters
Streaming architecture is real and valuable. The question is whether your use case actually requires it — and in regulated financial services, the answer is less often than the tooling conversation suggests.
Read →2026-05-10
Building Data Teams in Regulated Environments
Data team building in financial services requires a different profile than most hiring frameworks account for. The skills that matter most are the hardest to evaluate.
Read →2026-05-08
The Integration Problem Nobody Wants to Solve
Enterprise integration is unglamorous, expensive, and absolutely central to data quality in financial services. It's also the part of the architecture that gets deferred longest.
Read →2026-05-06
Why Data Mesh Doesn't Work the Way You Think
Data mesh solves real problems. But the organizations adopting it are mostly adopting the vocabulary, not the organizational change that makes it work.
Read →2026-05-04
The Real Cost of a Regulatory Submission Error
The penalty is the visible part. The real cost of a regulatory submission error is what it reveals about your data infrastructure — and what it demands to fix it.
Read →2026-05-02
dbt, Spark, or PL/SQL: Why the Tool Choice Is the Wrong Conversation
Every few years the industry produces a new consensus about the right data tooling. Organizations that reorganize around the tool rather than the problem pay for it.
Read →2026-04-30
The Myth of the Single Source of Truth
Every data strategy deck promises a single source of truth. Almost none of them explain what happens when the business has multiple legitimate truths that genuinely conflict.
Read →2026-04-28
What GDPR Actually Changed in Turkish Financial Services
KVKK didn't just add a compliance checkbox. For data teams in Turkish banks and insurance companies, it fundamentally changed what responsible data architecture looks like.
Read →2026-04-26
Why Your ETL Pipeline Is a Liability, Not an Asset
The pipeline that took six months to build and has been running reliably for three years is often the biggest risk on your data team's balance sheet.
Read →2026-04-24
The Data Debt Nobody Talks About
Technical debt gets board-level attention. Data debt accumulates invisibly — until the audit, the regulatory submission, or the executive dashboard shows something nobody can explain.
Read →2026-03-10
AI in Enterprise Data: The Conversation Nobody's Having
Everyone's talking about AI replacing analysts. Nobody's talking about what happens when your data quality is too poor for AI to actually use.
Read →2026-02-18
What I Actually Look for When Hiring Data Talent
After years of hiring in fintech and insurance, I've learned to ignore most CVs and ask a very specific set of questions instead.
Read →2026-01-22
Legacy Systems Aren't the Problem You Think They Are
The industry obsession with 'modernization' is causing organizations to destroy institutional knowledge they don't even know they have.
Read →2025-12-05
Regulatory Compliance Is an Architecture Problem
Most organizations treat compliance as a reporting exercise. The ones that don't get burned spend years rebuilding pipelines from scratch.
Read →2025-10-14
Data Governance: Why Most Companies Are Fooling Themselves
A governance framework that lives in a PowerPoint deck isn't governance. It's theater. Here's how to tell the difference.
Read →Also writing at Teknovole ↗