01 / Risk Observatory
Cross-domain investigation · evidence-led judgement
Investigation scope
AML · KYC · SANCTIONS · QA
Somesh Kumar JhaFinancial Crime ComplianceBengaluru · India
Investigate the signal.
Build the evidence.
Defend the decision.
Financial Crime work is rarely one domain at a time. A payment alert can become a customer-risk question; an ECDD review can expose transaction behaviour; adverse media can change the weight of an otherwise ordinary name match. My work sits where those signals converge.
RISK ATLAS / LIVEsignals converge before judgement
Identity
Payments
Ownership
Activity
External risk
FINCRIMEDECISIONevidence weighted · rationale documented
01Signalalert · customer · activity
02Evidencecontext · source · history
03Judgementrelease · escalate · investigate
AML / CFTKYC · CDD · ECDDTM & Transaction AnalysisSanctions & PaymentsAdverse Media & OSINTQA / QCSMR · UMR · CTR · STR/SAR context
02 / Investigation Lab
One investigation.
Multiple risk lenses.
Select a live work pattern. The evidence board, systems, risk questions and decision path change with it — this is not a one-word state switch.
LIVE PAYMENTsignal → context → decision
03 / Evidence Stack
What changes
the weight of a case.
Evidence is useful only when it changes a hypothesis, corroborates a risk factor, or makes the final rationale more defensible. Choose any evidence layer below.
04 / Systems I Operate
Tools are not badges.
They are evidence paths.
The value is not knowing a product name. It is knowing where to look, what to test, how to reconstruct context and what the system can — and cannot — prove.
05 / AI × FinCrime
Use AI to compress noise.
Never outsource judgement.
My interest in AI is operational: where can it shorten investigation time, surface connections and improve quality without turning uncertain evidence into automated certainty?
ALERT TRIAGEreal-time investigation model
AI can rank, retrieve, connect and summarise. The accountable FinCrime disposition stays with a trained analyst.
Outputs should point back to the underlying customer, payment, transaction or external source — not create opaque conclusions.
Automation should reduce repetitive noise while routing ambiguity, sanctions exposure and contradictory evidence to human review.
QA outcomes can become a control signal: identify repeat defects, improve prompts/rules, coach investigators and strengthen procedures.
06 / Career Dossier
Volume taught speed.
Complexity taught judgement.
My experience developed across high-volume sanctions operations and high-risk customer investigations, with quality, escalation and process-improvement thinking running through both.
SOCIÉTÉ GÉNÉRALESanctions Screening · Cross-border Payments
L1 / validator-oriented operating environmentReviewed live payment alerts across high-volume global queues, interrogating names, beneficiaries, counterparties, countries, structured payment fields and sanctions regimes before release or escalation. Historical context was reconstructed through Intix and Firco Archive rather than treating each hit as an isolated string match.
Firco Live / ContinuityFirco ArchiveIntixGlobal RelayOFAC / UN / EUSWIFTIBFSSEPAEBA / TGT
- Worked at high daily alert volumes while maintaining evidence-led disposition quality.
- Escalated cases where match quality, geography, payment context or counterparty evidence warranted deeper review.
- Contributed an automation/process-improvement idea that was implemented in the operating environment.
COMMONWEALTH BANK OF AUSTRALIAECDD · High-Risk Customer Investigation
Individuals + entities · enhanced due diligenceInvestigated complex ECDD cases spanning customer identity, beneficial ownership, Source of Funds, Source of Wealth, transaction behaviour, adverse media and connected-party risk. CommSee provided relationship-level evidence while SIRIUS anchored the case workflow and World-Check supported name-risk screening.
SIRIUSCommSeeLSEG World-CheckGenAI + manual AMUBO / BOSoF / SoWPEP / RCARE-KYCTM
- Completed and disposed 50+ complex ECDD cases in a 13-day high-output period.
- Reviewed high-risk individuals and entities, including ownership structures and customer-document evidence.
- Used GenAI selectively to accelerate adverse-media discovery, with manual corroboration before reliance.
07 / Contact
Bring me the
risk problem.
I’m interested in Financial Crime roles where investigation depth, payment risk, customer-risk analysis, quality and thoughtful control improvement matter — across AML, KYC/CDD/ECDD, TM, sanctions, fraud-risk adjacencies and QA/QC.