Position Summary We are seeking a hands-on MDM Specialist to take a custom, warehouse-native Customer MDM solution from design and partially implemented SQL to a reliable running master-data hub. The solution does not use a commercial MDM platform. Its party model, source crosswalks, matching and scoring logic, survivorship rules, reversible merge mechanics, and golden-record pipelines are implemented directly using Snowflake, dbt, SQL, and PostgreSQL.

The role focuses on Customer and Individual/Consumer mastering across D&B;, SAP, MTS, Sideline, DAL, and other source systems. The successful candidate will challenge and refine the existing design, write production code, measure match quality against labelled data, and iteratively execute the Load → Match → Merge → Review → Refine lifecycle until the resulting golden records are trustworthy, explainable, and traceable.

Important Role Context

This is a build role for a custom MDM hub—not a role configuring Informatica MDM, Reltio, Profisee, SAP MDG, Ataccama, or another packaged MDM product. Commercial MDM platform experience is not required and does not substitute for strong data modeling, record-linkage, SQL, and engineering skills. You will inspect the data, challenge assumptions, identify when documented processing is not actually running, and support conclusions with measured evidence.

Key Responsibilities 1. MDM Solution and Architecture Review Review the latest MDM solution design, including the architecture, Customer and Individual/Consumer data models, source systems, integration patterns, loading approach, and processing logic. Validate that the architecture supports initial and incremental loads, data quality controls, match and merge processing, exception handling, stewardship, lineage, auditability, restartability, and future scale.

Identify technical gaps, dependencies, implementation risks, and practical improvement opportunities.

- Customer and Consumer Data Modeling Validate the MDM data model for organization, customer, individual, consumer, household, account, affiliation, and hierarchy use cases. Assess support for identifiers, names, addresses, emails, phone numbers, classifications, roles, relationships, statuses, lifecycle dates, consent, privacy preferences, source lineage, and history. Identify missing attributes, entities, relationships, reference data, and metadata; propose and implement required logical and physical model changes. Ensure the model retains source-record cross-references, attribute-level provenance, match evidence, survivorship decisions, and golden-record history. Implement a party model in which organizations and people share a governed party root and “customer” is represented as a role, with appropriate subtypes, identifiers, relationships, lineage, and merge events. Model multiple hierarchies as governed edges—such as legal, operational, affiliation, and commercial rollups—so only authorized relationship types drive aggregation.
- Source-System Analysis and Mapping Profile and analyze data from Dun & Bradstreet (D&B;), SAP, MTS, Sideline, DAL, and other relevant source systems. Work with source-system SMEs to understand business definitions, ownership, quality, refresh frequency, identifiers, and downstream usage. Identify duplicate, incomplete, inconsistent, conflicting, stale, and invalid data and quantify their effect on mastering outcomes. Define source-to-MDM mappings covering direct mappings, transformations, standardization, reference translations, default and derived values, mandatory fields, data-quality validations, and rejection rules. Determine which MTS information is relevant to MDM and define how it should be incorporated into the model, match process, and golden record.
- Data Loading and Snowflake Development Design and prepare the initial data-loading process and support subsequent incremental and restartable loads. Review, develop, and optimize existing Snowflake SQL, scripts, stored procedures, views, queries, loading sequences, dependencies, and processing logic. Implement a controlled processing sequence for ingestion, staging, profiling, cleansing, standardization, validation, candidate generation, matching, scoring, merging, survivorship, golden-record creation, exception review, and publication. Build reconciliation controls, record-count checks, audit logging, error handling, recovery procedures, and operational monitoring.
- Matching, Scoring, and Entity Resolution Analyze and validate existing matching and scoring logic; identify false-positive and false-negative risks. Define and refine deterministic, probabilistic,



fuzzy, and composite rules for Customer and Individual/Consumer matching. Use relevant identifiers and attributes, including source customer IDs, D-U-N-S Numbers, names, standardized addresses, emails, phone numbers, tax or registration identifiers, geography, affiliations, and account relationships. Define blocking and candidate-selection strategies and calibrate thresholds for automatic match, possible match/manual review, and non-match outcomes. Generate transparent match scores, reason codes, evidence, and rule versions to support stewardship and audit review. Tune blocking recall, match precision, match recall, and thresholds against labelled links or other approved ground truth, reporting results separately for relevant consuming uses. Apply negative evidence and veto rules where shared or conflicting contact points would otherwise create unsafe merges. Resolve people at consumer scale where reference anchors may not exist and shared email addresses or phone numbers can connect multiple accounts. Use stricter person-match thresholds where over-merging could expose one consumer’s data to another, treating privacy risk differently from an organization-matching error.
- Survivorship, Merge, and Golden-Record Creation Define attribute-level survivorship rules based on source trust, quality, completeness, verification status, recency, consistency, and business-defined precedence. Progress the merge, cross-reference, and golden-record creation process while preserving source traceability and valid source detail. Support merge, unmerge, manual override, exception review, and controlled reprocessing requirements. Maintain an auditable explanation of why each golden value was selected and ensure matching and survivorship rules are version controlled. Implement crosswalk-based, reversible merges that never mutate original source records, record merge events, support replay/unmerge, and provide defined redirects for merged-away identifiers. Implement survivorship as governed, data-driven rules where practical, with firing telemetry, effective dating, and regression gates. Treat consent and communication preferences as source-specific governed assertions; do not combine them through ordinary survivorship logic. 7.

Iterative

Execution and Validation Execute the MDM lifecycle iteratively: Load → Standardize → Match → Merge → Review → Refine → Validate. Analyze each cycle and refine mappings, standardization logic, thresholds, matching rules, and survivorship rules based on evidence. Validate master and golden records for accuracy, completeness, uniqueness, consistency, usability, and source-to-golden traceability.

Investigate duplicate clusters, incorrect merges, missed matches, conflicting values, incomplete records, and other anomalies. Partner with business data stewards to resolve ambiguous matches and execute regression testing after rule or model changes.

- Documentation and Collaboration Document source-to-MDM mappings, data-model changes, cleansing rules, data-quality controls, match/scoring rules, thresholds, reason codes, survivorship rules, processing sequences, issues, and decisions. Maintain reconciliation results, validation evidence, exception logs, and technical runbooks. Collaborate with data architects, business analysts, data stewards, governance teams, source-system owners, QA teams, and downstream consumers. Translate business policies and data definitions into maintainable, testable, and explainable MDM logic.

Required Qualifications Bachelor’s degree in Computer Science, Information Systems, Engineering, Data Management, or a related discipline. 10+ years of experience in data engineering, data integration, or enterprise data management, including substantial hands-on responsibility for customer mastering or identity resolution. Robust party and entity-relationship modeling skills: ability to reason from first principles about entities, roles, identifiers, relationships, identity, lineage, and the consequences of different modeling choices. Hands-on probabilistic record-linkage experience covering blocking, candidate generation, scoring, thresholds, negative evidence, vetoes, and precision/recall trade-offs.





Experience tuning a matcher against labelled data or defensible ground truth and explaining the impact and cost of rule changes.

Experience with privacy-aware person matching at consumer scale, including ambiguous or heavily shared contact points and the asymmetric risk of over-merging individuals. Strong understanding of attribute-level survivorship, source provenance, mirrored-source risks, consent handling, and rule governance.

Experience designing or implementing reversible, crosswalk-based merge/unmerge processing with immutable source records and stable identifier resolution. Advanced SQL skills for large datasets, including window functions, recursive CTEs, closure tables, set-based MERGE patterns, query tuning, and warehouse-scale performance reasoning. Hands-on dbt or equivalent transformation-as-code experience, including tests, lineage, modular models, documentation, source freshness, and version control.

Experience reconciling an internal customer estate to authoritative external reference data and de-duplicating reference sources before they influence mastered identities. Strong ability to profile real data, validate what processing actually occurred, communicate discrepancies directly, and document evidence-based decisions.

Preferred Qualifications Hands-on Snowflake and PostgreSQL experience in a warehouse-native data hub or large-scale analytical environment.

Experience integrating and reconciling D&B; data and D-U-N-S identifiers with an internal customer estate. Knowledge of SAP customer-master structures and extraction patterns, plus experience with diverse operational source systems. Python experience for data profiling, transformation libraries, match evaluation, and repeatable quality analysis. Graph or network modeling experience for identity resolution, connected components, and hierarchy traversal.

Experience applying embeddings or vector similarity to entity resolution or attribute normalization with appropriate validation controls.

Experience designing data-stewardship workflows, exception queues, human-in-the-loop review, and feedback capture. Exposure to US education, K–12, youth-sports, institutional, or membership-oriented customer data is valuable.

Expected Deliverables

Reviewed solution design with a documented architecture, data-model, SQL/process, and implementation gap assessment. Validated party model for organizations and individuals, including governed relationship and hierarchy edges. Source-data profiling and quality assessment for D&B;, SAP, MTS, Sideline, DAL, and other approved sources.

Approved source-to-MDM mappings, reference-data reconciliation rules, and MTS incorporation approach. Production-ready initial and incremental loading models using Snowflake/dbt/PostgreSQL with testing, monitoring, reconciliation, and recovery controls. Measured match baseline and improved blocking recall, precision, recall, thresholds, negative evidence, and veto rules against approved ground truth.

Versioned, attribute-level survivorship rules with provenance controls, rule-firing telemetry, and regression tests. Reversible crosswalk-based merge/unmerge process and golden-record pipeline with lineage and merged-ID resolution. Golden-record validation results, exception analysis, issue/decision register, and prioritized remediation plan.

Technical documentation and operational runbook for repeatable Load → Match → Merge → Review → Refine execution.

Success Measures

Improved and measured blocking recall, match precision, and match recall against labelled data or approved ground truth. Privacy-safe person matching with explicit controls for shared contact points and over-merge risk. Accurate, complete, unique, explainable, and traceable golden customer records.

Reversible merges with immutable source records, complete crosswalks, and reliable merged-identifier resolution. Complete attribute-level provenance and defensible golden-value selection. Reconciled counts and tests across source, staging, match, merge, golden, and publish stages.

Repeatable, restartable, monitored, version-controlled, and auditable MDM processing. Business and data-steward approval of matching, survivorship, hierarchy, and golden-record outcomes. Core Technology and Domain Keywords Custom MDM Hub Party Model Customer 360 Consumer Identity Snowflake dbt PostgreSQL Advanced SQL D&B; D-U-N-S SAP MTS Sideline DAL Probabilistic Record Linkage Blocking Candidate Generation Precision/Recall Attribute-Level Survivorship Reversible Merge/Unmerge Crosswalk Golden Record Hierarchy Edges Lineage Stewardship Reconciliation

📌 Senior MDM Specialist (Bengaluru)
🏢 Varsity
📍 Bengaluru