Gunaditya checks data quality and reconciliation

Plain language meaning

Gunaditya's data quality and reconciliation check is the control gate where Malla Bank proves that the loan application, KYC, account, income, bureau and transaction-history data are complete, current, reconciled and usable before AI or ML scoring begins.

This topic is about banking data control in a lending pipeline. It is not about generic data cleansing or cosmetic dashboard quality.

In a real bank, this is not a loose technology idea. It is a controlled operating step where customer facts, banking policy, model behaviour, human authority, legal obligations and retained evidence must line up. AI and ML can improve speed, consistency and detection quality, but the bank must still prove why the process was fair, explainable, secure, monitored and fit for purpose.

Where it sits in the banking AI journey

This card belongs to The Complete Pipeline. The working flow is Source extract, Quality rules, Reconciliation, Exception queue, and Approved data set.

Read the flow as a banking control journey. Each stage needs a source system, a decision purpose, a failure mode, a control owner, a fallback path, a customer-impact view and retained evidence. Without those elements, the bank may have automation, but it does not yet have a bank-grade AI process.

Banking data and evidence

The important data points are application ID, customer master, KYC status, income amount, account balance, transaction count, bureau reference, and record total. These items matter because they influence lending eligibility, affordability, fraud risk, compliance treatment, operational queueing, regulatory reporting, customer explanation and audit traceability.

The evidence pack should include quality report, reconciliation result, exception register, source total, resolved defect note, approval timestamp, and data lineage record. A strong bank can replay the case from source data to feature values, model output, control result, human review, final outcome and monitoring result. A weak bank only knows that a system produced an answer.

Controls that make AI adoption safe

The core controls are mandatory-field check, format validation, source reconciliation, duplicate detection, freshness test, materiality threshold, and exception ownership. These controls make the topic bank-grade because they tie technical output to approved policy, legal obligations, model governance, operational resilience and management accountability.

AI can help compare records, detect anomalies, retrieve policy, summarise case evidence, prioritise work, highlight weak signals and improve investigator consistency. It should not invent missing facts, ignore failed checks, bypass authority, hide uncertainty, decide material customer outcomes without approval or create explanations that cannot be tied back to approved sources.

Regulatory and governance lens

For banking use cases, model risk, fair lending, adverse-action explanation, credit-risk governance, data lineage, operational resilience, AML/CFT risk-based controls, sanctions compliance, fraud information sharing and auditability can meet in the same workflow. The practical design must therefore be narrower and more disciplined than a generic AI design.

The practical test is simple: if a reviewer asks why the bank used the data, why the model output was trusted, why the customer received that action, why an alert was cleared, why an exception was approved, or why a regulatory record was prepared, the evidence must already exist.

Diagram walkthrough

Read the diagram from left to right as Source extract, Quality rules, Reconciliation, Exception queue, and Approved data set. The diagram is a control map, not decoration. It shows the minimum route by which data, AI or ML output, human action and audit evidence should connect.

Use it as a 30-minute study method. For each box, ask what system produces the data, what can go wrong, what control detects the weakness, who reviews the case, what customer or regulatory impact could arise and what record proves closure.

Most important mistake to avoid

The common failure is allowing a model to score because data exists, while the bank has not proved that the data is complete, reconciled and fit for the decision date.

The correction is to slow down the thinking, not necessarily the process. A well-designed banking AI process can be fast, but every fast step must still leave behind source lineage, control evidence, decision reason, human accountability, monitoring data and issue ownership.

Source anchors for accurate study

Federal Reserve SR 26-2, dated 17 April 2026, supersedes SR 11-7 and SR 21-8 for traditional model risk management and clarifies that generative and agentic AI need governance through broader risk-management controls.

NIST AI RMF 1.0 uses Govern, Map, Measure and Manage functions for AI risk management, and NIST AI 600-1 adds generative-AI risk actions for content provenance, hallucination, data protection, cybersecurity and human oversight.

U.S. Regulation B, 12 CFR 1002.9, requires specific principal reasons for adverse action in covered credit decisions, including when a creditor uses an AI model. CFPB Circular 2022-03 was withdrawn on 12 May 2025; do not cite it as current guidance. Primary sources: https://www.consumerfinance.gov/rules-policy/regulations/1002/9 and https://www.consumerfinance.gov/compliance/guidance/withdrawn-guidance/.

The EU AI Act treats AI systems used to evaluate creditworthiness or establish credit scores for natural persons as high-risk, except certain fraud detection and prudential capital contexts.

The Basel Framework IRB standards require banks to estimate and validate PD, LGD and EAD using relevant data, meaningful risk differentiation and ongoing governance.

FFIEC BSA/AML examination guidance expects suspicious activity monitoring systems to be risk-based, explainable by management, periodically reviewed and independently validated where appropriate.

Federal Reserve SR 26-3 and FinCEN's 12 June 2026 Section 314(b) materials clarify fraud-related information sharing under the USA PATRIOT Act safe-harbor framework for participating financial institutions.

OFAC's Framework for Compliance Commitments describes sanctions compliance programme components including management commitment, risk assessment, internal controls, testing and auditing, and training.

A source-to-decision reconciliation

Suppose the channel submits one loan application at 10:00 with verified monthly income, while the account feed shows a different amount because a reversal posted at 10:03. Gunaditya must first establish which record was available at the decision time and whether the application and account records describe the same customer, currency and period. A later correction is useful for investigation but must not silently replace the decision snapshot. The reconciliation should compare expected and received records, accepted and rejected rows, customer joins and the count of feature-ready applications. A green pipeline job only proves the job ran; it does not prove the right borrower was joined to the right evidence.

The exception register records the failed rule, affected application, owner, materiality, time detected, source correction and release decision. If a bureau file is late, the approved fallback may be referral rather than a score calculated from partial data. If an income record is duplicated, the bank should quarantine the derived feature, correct the lineage and assess whether any earlier decision used it. A test should deliberately add one duplicate, one stale record and one changed customer identifier, then verify that the decision service receives the expected block or referral. The audit sample must reconstruct both the first failed state and the corrected state.

Banking practice note: customer purpose

For gunaditya checks data quality and reconciliation, customer purpose is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from application ID to quality report. Then ask which control from mandatory-field check proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

AI can assist by comparing records, detecting unusual patterns, retrieving approved policy, summarising weak evidence, prioritising exceptions and preparing review notes. The bank should not allow a generated explanation, a confident score or a convenient dashboard to replace validation, consent, human judgement, customer communication, regulatory judgment or issue closure.

A strong implementation records the source event, data timestamp, consent or lawful basis, model or prompt version, feature values, score or generated output, threshold, reason code, user action, exception status, monitoring result, owner review and final outcome. That record lets risk, compliance, audit, technology and operations speak from the same facts.

Banking practice note: consent and lawful use

For gunaditya checks data quality and reconciliation, consent and lawful use is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from customer master to reconciliation result. Then ask which control from format validation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: source lineage

For gunaditya checks data quality and reconciliation, source lineage is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from KYC status to exception register. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: KYC and identity

For gunaditya checks data quality and reconciliation, KYC and identity is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from income amount to source total. Then ask which control from duplicate detection proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: account behaviour

For gunaditya checks data quality and reconciliation, account behaviour is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from account balance to resolved defect note. Then ask which control from freshness test proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: feature freshness

For gunaditya checks data quality and reconciliation, feature freshness is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from transaction count to approval timestamp. Then ask which control from materiality threshold proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: point-in-time correctness

For gunaditya checks data quality and reconciliation, point-in-time correctness is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from bureau reference to data lineage record. Then ask which control from exception ownership proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: model version

For gunaditya checks data quality and reconciliation, model version is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from record total to quality report. Then ask which control from mandatory-field check proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: decision threshold

For gunaditya checks data quality and reconciliation, decision threshold is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from application ID to reconciliation result. Then ask which control from format validation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: reason code

For gunaditya checks data quality and reconciliation, reason code is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from customer master to exception register. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: human review

For gunaditya checks data quality and reconciliation, human review is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from KYC status to source total. Then ask which control from duplicate detection proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: fraud control

For gunaditya checks data quality and reconciliation, fraud control is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from income amount to resolved defect note. Then ask which control from freshness test proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: sanctions control

For gunaditya checks data quality and reconciliation, sanctions control is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from account balance to approval timestamp. Then ask which control from materiality threshold proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: AML control

For gunaditya checks data quality and reconciliation, AML control is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from transaction count to data lineage record. Then ask which control from exception ownership proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: fair lending

For gunaditya checks data quality and reconciliation, fair lending is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from bureau reference to quality report. Then ask which control from mandatory-field check proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: regulatory reporting

For gunaditya checks data quality and reconciliation, regulatory reporting is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from record total to reconciliation result. Then ask which control from format validation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: operational exception

For gunaditya checks data quality and reconciliation, operational exception is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from application ID to exception register. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: customer harm

For gunaditya checks data quality and reconciliation, customer harm is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from customer master to source total. Then ask which control from duplicate detection proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: audit trail

For gunaditya checks data quality and reconciliation, audit trail is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from KYC status to resolved defect note. Then ask which control from freshness test proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: data quality

For gunaditya checks data quality and reconciliation, data quality is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from income amount to approval timestamp. Then ask which control from materiality threshold proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: privacy minimisation

For gunaditya checks data quality and reconciliation, privacy minimisation is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from account balance to data lineage record. Then ask which control from exception ownership proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: committee reporting

For gunaditya checks data quality and reconciliation, committee reporting is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from transaction count to quality report. Then ask which control from mandatory-field check proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: reconciliation

For gunaditya checks data quality and reconciliation, reconciliation is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from bureau reference to reconciliation result. Then ask which control from format validation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: exception ownership

For gunaditya checks data quality and reconciliation, exception ownership is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from record total to exception register. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: monitoring cadence

For gunaditya checks data quality and reconciliation, monitoring cadence is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from application ID to source total. Then ask which control from duplicate detection proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: closure evidence

For gunaditya checks data quality and reconciliation, closure evidence is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from customer master to resolved defect note. Then ask which control from freshness test proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: policy retrieval

For gunaditya checks data quality and reconciliation, policy retrieval is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from KYC status to approval timestamp. Then ask which control from materiality threshold proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: investigator feedback

For gunaditya checks data quality and reconciliation, investigator feedback is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from income amount to data lineage record. Then ask which control from exception ownership proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: model drift

For gunaditya checks data quality and reconciliation, model drift is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from account balance to quality report. Then ask which control from mandatory-field check proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: control attestation

For gunaditya checks data quality and reconciliation, control attestation is not a side detail. It decides whether the bank can connect the AI or ML output to a real banking purpose, a real customer or regulatory outcome and a real accountable owner. Study the topic as a banking process first and a model process second.

Trace one item from transaction count to reconciliation result. Then ask which control from format validation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: customer purpose

Trace one item from bureau reference to exception register. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: consent and lawful use

Trace one item from record total to source total. Then ask which control from duplicate detection proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: source lineage

Trace one item from application ID to resolved defect note. Then ask which control from freshness test proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: KYC and identity

Trace one item from customer master to approval timestamp. Then ask which control from materiality threshold proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: account behaviour

Trace one item from KYC status to data lineage record. Then ask which control from exception ownership proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: feature freshness

Trace one item from income amount to quality report. Then ask which control from mandatory-field check proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: point-in-time correctness

Trace one item from account balance to reconciliation result. Then ask which control from format validation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: model version

Trace one item from transaction count to exception register. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: decision threshold

Trace one item from bureau reference to source total. Then ask which control from duplicate detection proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: reason code

Trace one item from record total to resolved defect note. Then ask which control from freshness test proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: human review

Trace one item from application ID to approval timestamp. Then ask which control from materiality threshold proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: fraud control

Trace one item from customer master to data lineage record. Then ask which control from exception ownership proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: sanctions control

Trace one item from KYC status to quality report. Then ask which control from mandatory-field check proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: AML control

Trace one item from income amount to reconciliation result. Then ask which control from format validation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Banking practice note: fair lending

Trace one item from account balance to exception register. Then ask which control from source reconciliation proves the item was complete, current, authorised, relevant and fit for use. If that trace cannot be shown without manual guessing, the process is not yet bank-grade.

Show the exception before the model

Gunaditya receives a dated batch of 1,000 applications. The source manifest lists IDs and requested amounts; the feature join produces 998 rows. He identifies two missing applications, their source channels and balances before approving any score run. One may have an unresolved customer mapping; the other may be a duplicate submission. A 99.8% completeness figure alone does not justify silently excluding either.

He checks salary unit, account status, feature freshness and label separation on sampled rows. A recent salary correction can improve a later dataset but must not overwrite an earlier decision snapshot. If a critical field is invalid, he assigns a source owner and records containment, repair and rerun IDs. Downstream consumers see an explicit incomplete state. Closure requires reconciling both the dataset and final customer actions, not merely clearing a technical queue.