Post Job Free
Sign in

Data Quality & Governance Rules-Based Automation

Location:
Pompton Plains, NJ
Posted:
November 18, 2026

Contact this candidate

Resume:

RAJENDER DICHPALLY

Senior Data Management Professional — Data Quality and Governance, Rules-Based Automation, Production Python, Exception Workflows

Parsippany, NJ 973-***-**** ********.*********@*****.***

U.S. Citizen

PROFESSIONAL SUMMARY

●24 years in data quality management, data operations, data acquisition and data governance inside the finance industry — well past the three years required — across Bank of America, Mizuho Bank, JP Morgan Chase, Nomura Securities, Standard & Poor's, PricewaterhouseCoopers, Hudson Insurance Group and Ambac Assurance.

●Data quality is the specialism rather than a responsibility within another role: quality rules defined and implemented, measured, monitored and enforced at the point data enters a system rather than discovered downstream in a report.

●Rules-based automation applied to acquisition, enrichment, validation and operational workflows throughout — detection and business rules on Bank of America's surveillance platform, compliance rule logic on PricewaterhouseCoopers' Global Independence platform, and validation and reconciliation rulesets on statutory reporting at Hudson Insurance Group.

●Governance and holdings data at PricewaterhouseCoopers: I delivered the firm-wide Global Independence Compliance platform monitoring the security holdings of every partner and manager against compliance rules, with the rule logic, data sources and reporting outputs specified by me and a team of eight developers led remotely.

●That platform ingested market data feeds from Bloomberg and Reuters through procedures I wrote — acquiring, normalizing and matching external reference data against an internal population of individuals, which is the same problem shape as onboarding and maintaining an entity-and-people dataset.

●Identifying recurring manual processes and converting them into automation is a documented habit across four firms — manual, spreadsheet-heavy processes replaced with documented, automated, reconciled ones at Hudson Insurance Group, JP Morgan Chase and The Vitamin Shoppe; report generation and distribution, archiving and purging and replication monitoring automated at JP Morgan Chase; environment deployment, configuration and housekeeping automated at United Nations UNICEF; and application and database deployment automated at PricewaterhouseCoopers.

●The principle I work to: the second time an issue is resolved by hand, the fix is to remove the cause or automate the response — and a third occurrence of the same ticket is an engineering failure rather than an operations one.

●Preventive controls rather than detective ones wherever possible — a quality failure caught at acquisition costs minutes, the same failure caught in a client-facing dataset costs trust.

●Production Python for data workflows and analytical tooling: extraction from REST APIs with authentication, pagination, rate handling and retry, and from databases, files and spreadsheets; classification, parsing and enrichment; profiling and validation; reconciliation between independent sources; exception reporting; and scheduled execution with argument handling, error trapping, logging, alerting and retry — written as reusable parameterized modules that other people run.

●Data profiling as the first move on any new source — distribution, frequency, nulls, blanks and placeholder values distinguished from each other, cardinality, uniqueness and referential integrity, value domains against what the business believes they are, pattern and format consistency, duplicate and near-duplicate detection, orphan and gap analysis, and outliers against expected ranges.

●Data quality metrics and monitoring frameworks built rather than described: completeness, validity, consistency, uniqueness, timeliness and accuracy measured per source and per field, thresholds set from observed behaviour rather than round numbers, and anomaly detection against expected volumes and distributions so a degraded or truncated feed surfaces before a consumer notices.

●Human-in-the-loop design as standing practice — exceptions captured, triaged and routed to whoever can actually fix the record rather than to a queue nobody owns; review and approval gates where the stakes justify them; and the output of the review fed back into the rules so the same exception stops recurring.

●Entity resolution and matching across systems: the authoritative source established per field, the points where two systems disagree made explicit rather than averaged away, matching rules defined with the handling of non-matches stated, and duplicate and near-duplicate detection across inconsistent naming and formatting.

●Data pipelines and workflow orchestration at enterprise scale — SSIS and Informatica PowerCenter with error output routing to reject tables with reason codes, checkpoints and restartability; Autosys, SQL Server Agent, Oracle Scheduler and cron for dependency chains with conditional branching, retry and escalation; incremental and delta loading with watermark design; idempotent reloads; and restart and recovery from the point of failure.

●Success measured rather than asserted: the effect of an automation or rule change estimated before it is made, measured after it, and the difference explained — with manual effort removed, exception volume, coverage and quality tracked as the metrics that matter.

●SQL at expert level across SQL Server, Oracle and Exadata, PostgreSQL, Sybase, DB2 and MySQL, used daily to profile a source, prove a population and validate what a system is actually doing rather than what it is documented to do.

●Educating colleagues on data quality and scalable practice through code review with reasons attached, design walkthroughs before the build, and standards established at PricewaterhouseCoopers and Bank of America where none previously existed — plus runbooks and templates left behind so capability outlasts the engagement.

●Tableau across five consecutive engagements with Power BI and SSRS alongside, used for quality dashboards and operational monitoring as well as business reporting.

●M.S. Computer Science, New York Institute of Technology, with an MBA in Finance. U.S. Citizen. Parsippany, NJ — a straightforward commute into New York City. Available immediately.

REQUIREMENTS SNAPSHOT

3+ years in Data Quality Management, Data Management, Data Operations, Data Acquisition or Data Governance within Finance or Technology — required: Twenty-four years, almost entirely within finance. Data quality management: quality rules defined, implemented, measured and monitored across every engagement, with reconciliation to the penny against independent finance and actuarial sources at Hudson Insurance Group and root-cause analysis on every discrepancy. Data operations: statutory, regulatory and statistical reporting run end to end against immovable filing deadlines across multiple lines, states and legal entities. Data acquisition: heterogeneous source onboarding across six database platforms, REST APIs, flat and fixed-width files, XML and JSON, mainframe extracts, and market data feeds from Bloomberg and Reuters. Data governance: definitions and business glossary, lineage, stewardship named per domain, and role-based access with row-level security, delivered inside control environments examined by internal audit, external audit and state regulators.

Experience applying AI, machine learning, NLP or rules-based automation to data acquisition, enrichment, validation or operational workflows — required: Rules-based automation is where my record is strong, applied to all four of those areas. Acquisition: feeds triggered on file arrival with dependency handling at Nomura Securities so downstream processing started only once expected files had landed complete; market data feeds from Bloomberg and Reuters ingested through procedures I wrote at PricewaterhouseCoopers. Enrichment: reference-data lookup, derivation and standardization rules applied in the conformed layer once and reused rather than re-implemented per consumer. Validation: record count, control total, header and trailer checks; referential integrity, uniqueness, domain and range rules per field; threshold, tolerance and variance checks; anomaly detection against expected volumes and distributions. Operational workflows: detection and business rules on Bank of America's surveillance platform specified, implemented and tuned, with the rationale and expected behaviour documented for every rule so output could be validated against intent; and compliance rule logic on PricewaterhouseCoopers' Global Independence platform. Machine learning, NLP and AI tooling specifically are not in my record.

People, governance and company data — we'd love to see: The closest work in my record is a genuine match in shape. At PricewaterhouseCoopers I delivered the Global Independence Compliance platform, which monitored the security holdings of every partner and manager across the firm against compliance rules — an entity-and-people dataset held against external market reference data, with the rule logic, data sources and reporting outputs specified by me and a team of eight developers led remotely. That meant acquiring external market data feeds from Bloomberg and Reuters, normalizing them, matching securities and individuals across systems with inconsistent identifiers, applying rules to detect breaches, and routing exceptions to compliance for review. Alongside it: ratings and index data products at Standard & Poor's; corporate and counterparty reference data across Mizuho Bank, Nomura Securities and Bank of America; and financial data throughout.

Identify recurring manual processes, bottlenecks and quality failure points, and translate them into automation and preventive controls — core responsibility: A documented habit across four firms rather than an initiative. Manual, spreadsheet-heavy processes replaced with documented, automated, reconciled ones at Hudson Insurance Group, JP Morgan Chase and The Vitamin Shoppe. Report generation and distribution, data archiving and purging and replication monitoring automated at JP Morgan Chase. Database environment deployment, configuration and housekeeping automated at United Nations UNICEF. Application and database deployment automated at PricewaterhouseCoopers. The method: watch what people actually do by hand rather than wait for a request, because the workaround nobody documented is where the recurring cost lives; establish whether the manual step exists because the process needs judgment or because the system never handled a case; and prefer removing the cause over automating the symptom. Preventive over detective wherever possible — a quality failure caught at acquisition costs minutes, the same failure caught in a client-facing dataset costs trust.

Design, build and deploy automation that improves data quality, expands coverage and scales onboarding and maintenance — core responsibility: Built in Python, SQL, SSIS and shell, orchestrated through Autosys, SQL Server Agent, Oracle Scheduler and cron with dependency chains, conditional branching, retry and escalation. Scaling onboarding specifically: source-to-target mapping templates so every new source documents the same things; parameterized, reusable modules rather than one-off scripts per feed; a profiling and validation pattern applied to every source before it is trusted; error routing to reject tables with reason codes so bad records are captured rather than failing a run; incremental and delta loading with watermark and control-table design; idempotent reloads so a rerun is safe; and restart and recovery from the point of failure. One worked pattern documented properly so the next twenty follow it, which is what actually makes onboarding scale.

Perform data profiling and apply analytical methods to support quality measurements, rulesets and monitoring frameworks — core responsibility: Profiling first, every time: distribution, frequency, nulls, blanks and placeholder values distinguished from each other; cardinality, uniqueness and referential integrity; value domains tested against what the business believes they are; pattern and format consistency; duplicate and near-duplicate detection; orphan and gap analysis across keys; and outliers against expected ranges. Then turned into measurement rather than left as a finding: completeness, validity, consistency, uniqueness, timeliness and accuracy measured per source and per field; thresholds set from observed behaviour in comparable periods rather than round numbers; rulesets expressed so a failure identifies the record and the reason rather than only the count; and monitoring that alerts on deviation from baseline and on silence as well as on failure, because a feed that stops arriving produces no errors at all.

Design and implement workflows to classify, extract, enrich, reconcile, validate and prioritize data at scale — core responsibility: All six, built as rules-based pipelines. Classification and extraction: parsing and shredding of XML, XSD, JSON, flat and fixed-width files with header and trailer validation, and extraction from REST APIs and six database platforms. Enrichment: reference-data lookup, derivation and standardization applied once in the conformed layer and reused. Reconciliation: population completeness proved before comparison, counts and control totals before balances, timing and classification differences separated from genuine errors because each has a different remedy, and every variance traced to the record that caused it. Validation: integrity, uniqueness, domain, range, threshold and variance rules per field. Prioritization: exceptions ranked by materiality and by whether they block a downstream deadline, so the queue reflects what actually matters rather than arrival order.

Human-in-the-loop workflows for data review, exception handling and quality control — we'd love to see: Standing practice rather than an occasional design choice, and in my view the part that determines whether automation is trusted. Exceptions captured with the evidence attached rather than as a bare flag, so the reviewer can act without going back to the data. Routed to whoever can actually correct the record rather than to a queue nobody owns. Triaged by materiality and by whether they block a deadline. Review and approval gates where the stakes justify them — at United Nations UNICEF, production changes were rolled in scheduled batches under documented review and approval; at Hudson Insurance Group nothing is filed until reconciliation and sign-off are complete. And the loop closed: the outcome of the review fed back into the rules, so a recurring exception becomes a rule change rather than a permanent queue. On Bank of America's surveillance platform this was the daily work — investigating alerts that proved unfounded, separating a threshold problem from a rule logic problem from a data quality defect, and tuning on the evidence.

Develop data quality metrics and business rules as part of a broader architecture or operational framework — required: Metrics tied to the architecture rather than reported beside it: measured at the layer where the defect originates, so completeness is checked at acquisition, conformance at the transformation layer and reconciliation at the consumption boundary. Business rules defined with the rationale recorded, the calculation stated unambiguously and the edge cases named — because a rule clear to its author and ambiguous to a developer is the most expensive kind of requirement. Applied once in a conformed layer and reused rather than re-implemented per consumer, which is what keeps two reports from disagreeing. Specified in this form on the Bank of America surveillance platform, the PricewaterhouseCoopers independence platform and statutory reporting at Hudson Insurance Group.

Understanding of data pipelines, workflow orchestration and automation frameworks at scale — required: SSIS across the full control and data flow transformation set with package configurations and parameterization, logging providers, event handlers, error output routing with reject tables and reason codes, checkpoints and restartability, and environment migration; Informatica PowerCenter; SQL*Loader and external tables; Python and Unix shell pipelines. Orchestration through Autosys for enterprise cross-system dependency chains, SQL Server Agent and Oracle Scheduler for database-resident scheduling, and cron — with conditional branching on upstream outcomes, parallel and sequential execution, task-level retry and failure escalation, backfill and catch-up runs, SLA monitoring against delivery windows, and alerting routed so the right person hears first.

Writing production-ready code and conducting unit and integration testing — required: Production code is what I write: stored procedures, functions and views running against live systems with real deadlines, and Python modules that other people run unattended. Written to be operable rather than merely correct — parameterized rather than hard-coded, instrumented with logging so a production run can be diagnosed without reproducing it, error handling that captures and routes bad records rather than failing the run, and idempotent so a rerun is safe. Testing: unit and integration testing of pipeline components; test plans, cases and artefacts maintained as deliverables; UAT coordinated with business users through scenario definition, execution tracking, defect triage to root cause, retest and sign-off; and every rewrite validated for identical output before acceptance, because a faster component that returns different numbers is a defect rather than an improvement.

Strong Python with production-grade data workflows, automation and analytical tooling — required: Python for the work this role describes. Acquisition: REST and web service API extraction with authentication, pagination, rate handling, error trapping and retry; database, file and spreadsheet extraction. Processing: parsing, classification, transformation, enrichment and standardization. Quality: profiling, validation, integrity checking, duplicate and near-duplicate detection, reconciliation between independent sources, and exception reporting with evidence attached. Operations: scheduled execution with argument handling, logging, alerting and retry; comparison harnesses for old-versus-new output during a change. All written as reusable parameterized modules rather than one-off scripts, because a script only its author can run is a liability rather than an asset. Alongside it, Unix shell, Perl and PowerShell for environment and housekeeping automation.

Define success metrics for automation impact, data quality improvement, operational efficiency and coverage — core responsibility: Measured rather than asserted, and the discipline is the same one I apply to any change: estimate the expected effect before making it, measure the actual effect afterwards, and explain the difference. Manual effort removed quantified in hours per cycle rather than described as a saving. Exception volume and the proportion resolving as genuine issues tracked, because falling exception counts can mean better data or a broken rule. Quality measured per source and per field against baseline. Coverage tracked as records and fields successfully onboarded and maintained. Timeliness against delivery windows. On Bank of America's surveillance platform, every threshold change was estimated, measured and explained in exactly this way.

Investigate complex operational challenges and independently develop scalable solutions — core responsibility: Independence is how my current engagement runs — sole analyst responsible for statutory and regulatory reporting end to end, from requirements through reconciliation to filing. Investigation method: establish what changed and when before forming a theory; separate a data problem from a logic problem from a process problem from an upstream problem, because each has a different owner and a different remedy; check completeness first, since a figure that moves because a feed arrived short is the most common false alarm; trace to the record that caused it rather than to the category it falls in; and treat 'why was this not caught earlier' as a finding in its own right. Then close it with a change to the rule, the pipeline, the validation or the monitoring rather than a workaround that defers it.

Collaborate with domain experts, Product and Engineering — core responsibility: Credible on both sides, which is what makes the collaboration work. With domain experts: their vocabulary used rather than mine, and the question behind the request established before any work starts. With engineering: requirements specified precisely enough to build from, with edge cases, exception paths and data realities stated rather than discovered in testing. With product: trade-offs framed in terms of cost, coverage and what a decision forecloses. Across 24 years with underwriters, controllers, finance, actuarial, compliance, risk, operations, trading desks, QA, DBAs, infrastructure and security teams, internal and external auditors, state regulators and executive management.

Educate and empower colleagues on data quality, automation and scalable data management — core responsibility: Teams of eight developers led entirely remotely at PricewaterhouseCoopers and four at Bank of America, both while personally building. Code review with reasons attached so the standard is learned rather than enforced; pairing; design walkthroughs before the build rather than review after it. Design, coding and documentation standards established and enforced at PricewaterhouseCoopers and Bank of America where none previously existed, adopted through review rather than decree. Runbooks, templates, documented procedures and parameterized modules left behind so colleagues work independently — the measure of success being how rarely I am called afterwards.

Agile and Scrum — we'd love to see: Agile and Scrum as a pod member alongside engineers, QA and product: user stories with acceptance criteria written so a tester can act on them, backlog refinement and estimation, sprint planning and daily stand-ups, demos, retrospectives acted on, JIRA for tracking and Confluence for documentation and decision records — alongside waterfall and hybrid governance where a regulatory milestone sets the date.

Tableau — we'd love to see: Tableau as primary visualization tool across five consecutive engagements — dashboards and metrics views, governed published data sources and extracts with scheduled refresh, calculated fields and level-of-detail expressions, parameters, filters, sets and actions, drill-down from the headline metric to the individual record, row-level security, projects, permissions and subscriptions. Used for data quality dashboards and operational monitoring as well as business reporting. Power BI and SSRS alongside. QlikSense specifically is not in my record.

Education, location, availability and authorization — required: M.S. Computer Science, New York Institute of Technology, New York, NY — exceeding the bachelor's requirement in Computer Science, Mathematics, Information Systems or Finance. Also MBA Finance and MBA Marketing. Based in Parsippany, NJ, a straightforward commute into New York City. Available to start immediately. U.S. Citizen — no sponsorship required now or in the future.

TECHNICAL SKILLS

Data Quality Management: Quality rules defined, implemented, measured and enforced at acquisition rather than discovered downstream; completeness, validity, consistency, uniqueness, timeliness and accuracy measured per source and per field; thresholds set from observed behaviour rather than round numbers; rulesets that identify the record and the reason rather than only the count; anomaly detection against expected volumes and distributions; monitoring that alerts on deviation from baseline and on silence as well as failure; reconciliation to the penny against independent sources with every variance root-caused

Data Profiling: Distribution, frequency, nulls, blanks and placeholder values distinguished; cardinality, uniqueness and referential integrity; value domains tested against business belief; pattern and format consistency; duplicate and near-duplicate detection; orphan and gap analysis across keys; outliers against expected ranges; trend and seasonality across reporting periods; findings turned into measurement and rules rather than left as observations

Rules-Based Automation: Detection and business rules specified, implemented and tuned on Bank of America's surveillance platform with rationale and expected behaviour documented per rule; compliance rule logic on PricewaterhouseCoopers' Global Independence platform; validation and reconciliation rulesets on statutory reporting; reference-data lookup, derivation and standardization rules applied once in a conformed layer and reused; file-arrival triggering with dependency handling; error routing to reject tables with reason codes

Human-in-the-Loop & Exception Handling: Exceptions captured with evidence attached rather than as a bare flag; routed to whoever can correct the record rather than an unowned queue; triaged by materiality and by whether they block a deadline; review and approval gates where stakes justify them, including scheduled production batches under documented review at United Nations UNICEF and sign-off before filing at Hudson Insurance Group; review outcomes fed back into the rules so recurring exceptions become rule changes; alert investigation and threshold tuning on the Bank of America surveillance platform

Python & Production Tooling: REST API extraction with authentication, pagination, rate handling, error trapping and retry; database, file and spreadsheet extraction; parsing, classification, transformation, enrichment and standardization; profiling, validation and integrity checking; duplicate and near-duplicate detection; reconciliation between independent sources; exception reporting with evidence; scheduled execution with argument handling, logging, alerting and retry; old-versus-new comparison harnesses; reusable parameterized modules; Unix shell, Perl, PowerShell

Pipelines & Orchestration: SSIS — full control and data flow transformation set, package configurations and parameterization, logging providers, event handlers, error output routing with reject tables and reason codes, checkpoints and restartability, environment migration; Informatica PowerCenter; SQL*Loader and external tables; incremental and delta loading with watermark and control-table design; late-arriving record handling; idempotent reloads; restart and recovery from point of failure; Autosys, SQL Server Agent, Oracle Scheduler and cron with dependency chains, conditional branching, retry, escalation, backfill and SLA monitoring

Data Acquisition & Onboarding: Heterogeneous source onboarding across six database platforms; REST and web service APIs; flat and fixed-width files with header and trailer validation; XML, XSD, XQuery and JSON parsing and shredding; Excel and CSV with schema drift handling; mainframe extracts; market data feeds from Bloomberg and Reuters ingested at PricewaterhouseCoopers; Linked Servers and OPENQUERY federation; source-to-target mapping templates so every new source documents the same things

Entity Resolution & Reference Data: Authoritative source established per field with disagreement points made explicit; matching rules defined across systems with inconsistent identifiers and the handling of non-matches stated; duplicate and near-duplicate detection across inconsistent naming and formatting; reference data managed as a first-class dependency; securities and individuals matched against external market data on the PwC independence platform

Data Governance: Definitions and business glossary; quality rules at capture and at load; lineage documented from source through transformation to consumption; metadata captured during the build; ownership and stewardship named per domain; role-based access control and least privilege; row-level and object-level security; masking and de-identification for non-production; segregation of duties; periodic access review; retention and archival; audit evidence produced for internal audit, external audit and state regulators

SQL & Databases: SQL Server 2014/2012/2008 R2, Oracle 12c/11g/10g/9i, Oracle Exadata, PostgreSQL, Sybase, DB2, MySQL; T-SQL and PL/SQL; complex multi-source joins, aggregations with ROLLUP and CUBE, window and analytic functions, CTEs and recursive queries, set operations, MERGE, pivot and unpivot, dynamic SQL; stored procedures, functions, triggers and views; execution-plan analysis, optimizer statistics, index strategy, partitioning and tuning; data modelling with grain and key design, slowly changing dimensions and effective-dating

Testing & Delivery: Unit and integration testing of pipeline components; test plans, cases and artefacts maintained as deliverables; UAT coordination with scenario definition, execution tracking, defect triage to root cause, retest and sign-off; every rewrite validated for identical output before acceptance; Agile and Scrum with user stories, acceptance criteria, backlog refinement, sprint ceremonies, JIRA and Confluence; waterfall and hybrid where a regulatory milestone governs; change control with documented review and approval

Analytics, Visualization & Domain: Tableau across five consecutive engagements — dashboards, governed published data sources and extracts with scheduled refresh, calculated fields and LOD expressions, parameters, filters and actions, drill-down to the individual record, row-level security; Power BI and SSRS; quality dashboards and operational monitoring; descriptive statistics, distribution and variance analysis, cohort and period-over-period comparison, correlation and



Contact this candidate