Achyuth RachurFinancial services & technologyHome
Menu

Achyuth Rachur / Financial services × AI & operations

Complex work.Clear decisions.Better systems.

I work with clients to understand the workflow, data, and controls behind a problem—then design the right change through AI, automation, or process redesign and help carry it through delivery.

Explore selected work View résumé

Financial-services consulting · AI solution design · Workflow transformation · Chicago

Operator’s notebook01 / Working sketch
How does the work
actually happen?
PeopleDataRisk & controls
Process
redesignDeterministic
automationHuman-led
AI workflow
A testable change.
A clear human handoff.
Understand first. Choose the intervention second.
Conceptual illustration / not a client system.

From professional experience

Read the résumé context
85%
less Auditor-in-Charge timeCDD audit workflow redesign for a top-five U.S. financial institution.
~90%
faster GL reconciliation$300M in previously unresolved variance eliminated over three years for a community bank and fintech partner.
25+
client AI strategy sessionsStrategy and enablement delivery, alongside 15+ customized GenAI demos and technical walkthroughs.

The first two figures are prior engagement-reported outcomes; public baselines and attribution methods are not disclosed. Sessions and demos are delivery activity. These figures are separate from the independent portfolio examples.

01 / Selected work

Not just what I built.
Why each decision matters.

Two independently owned software projects and three fictional design exercises. Inspect the work, its decision boundaries, and the limits of its evidence.

Source lineage / real local artifact
SourceFictional Markdown
TransformPassthrough backend
ManifestSHA-256 retained
11 chunksSource + section + page
chunk / 0009Evidence rulespage:2
Reviewed CLI output11 chunks
Sourcefictional-expense-review-standard.mdSHA-256 22611878d966…
Generated chunk / 0009Evidence rulesFictional Expense Review Standard / Evidence rules

Source identity and section path stay attached to the converted text.

Real CLI output; invented source; static browser replay.

01 / Document AIIndependent software

Document Context

Prepare documents without losing their source.

Local conversion produces structured Markdown and source-linked chunks for downstream AI workflows.

Evidence: Real CLI artifact from a fictional standard

Design decisionKeep source identity and section context attached to each chunk.
Prototype · Tested artifactRead case
Proposed operating model
One request.
Connected decisions.
  1. 01Request
  2. 02Decompose
  3. 03Specialist challenges
  4. 04Human review
  5. 05Future state
  6. 06Change loop
Fictional design recordReview / 03
Use caseStaff knowledge assistant
Open decisionWhich local policy and approval path apply in a real organization?

Specialists receive questions; a model does not issue an approval.

Concept diagram; simulated prewritten responses; no running agents.

02 / AI operating modelFictional design concept

AI Use Case Studio

Make an AI use case reviewable before build.

A proposed workspace connects intake, dependencies, specialist questions, and future-state design.

Evidence: Fictional concept walkthrough

Design decisionKeep agent proposals separate from human approval.
Concept · IllustrativeRead case
Fictional working example
01 / SignalPossible duplicate
02 / Supporting recordCompare the entries
03 / Human ownerDecide the next check

Advisory queue · no automatic posting

Fictional journal reviewMay 2026
J-401Account 6150 / EAST$4,200
J-402Account 6150 / EAST$4,200

A possible duplicate needs evidence. No posting changes automatically.

Invented records; inspectable rules; human decisions.

05 / Financial closeIndependent prototype

Close Workbench

Bring close exceptions into one reviewer queue.

Journal and invoice signals appear with supporting records and the next human check.

Evidence: Working static prototype with fictional data

Design decisionTreat every flag as advisory, not an accounting conclusion.
Prototype · Tested artifactRead case

Two more decisions to inspect

Evidence ReviewFictional design exercise / Risk operationsShow missing evidence before a reviewer decides.Ledger MatchFictional design exercise / ReconciliationKeep ambiguous ledger matches open for review.
See all 5 projects

02 / What I actually do

The answer isn’t
always AI.

Start with the operation. Make the constraints visible. Choose an intervention the people doing the work can use.

  1. 01 / Fieldwork

    Understand
    the operation.

    Trace decisions, data, handoffs, exceptions, and controls with the people who know the work.

    Working output / current-state map
  2. 02 / Judgment

    Design
    the change.

    Compare process redesign, automation, and AI. Make the tradeoffs, ownership, and review points explicit.

    Working output / options + future state
  3. 03 / Delivery

    Make
    it real.

    Build a bounded example, test the difficult paths, and work through implementation, evaluation, and adoption.

    Working output / testable prototype + handoff
How I work with clients

03 / Professional experience

Experience in real operations.
A separate body of evidence.

These anonymized snapshots summarize cleared résumé outcomes. The independent examples above do not reproduce or validate these client implementations.

Financial crime / audit transformation

A more efficient
CDD audit workflow.

Starting problem
Customer-due-diligence audit effort at a top-five U.S. financial institution.
My contribution
Redesigned the audit workflow with agentic AI, as described in my cleared résumé.

85%less Auditor-in-Charge time; approximately 3,000 hours of annual financial-crime audit effort eliminated.

Engagement-reported outcome. Public baselines and attribution methods are not disclosed; no client artifacts are published here.

Banking operations / reconciliation

Resolve the variance.
Improve the process.

Starting problem
Previously unresolved general-ledger variance for a community bank and fintech partner.
My contribution
Developed the reconciliation program summarized in my cleared résumé.

~90%faster GL reconciliation; $300M in previously unresolved variance eliminated over three years.

Engagement-reported outcome. Public baselines and attribution methods are not disclosed; this is separate from Ledger Match.

Read professional experience
Portrait of Achyuth Rachur

04 / A little about me

A consultant’s questions.
A builder’s follow-through.

I began in model risk, learning to examine the data, controls, and decisions around a system. My work has since taken me across financial crime, audit, lending, banking operations, insurance, and AI solution design. I often enter as a learner: listen to the people doing the work, trace the underlying process, and build enough domain fluency to ask better questions. I connect business, risk, and engineering perspectives—and like staying with a solution through architecture, a bounded prototype, evaluation, delivery, and adoption. The projects here make a few of those decisions inspectable.

More about me

Let’s talk / Chicago

Have a difficult workflow
worth fixing?

Working on financial-services operations or AI transformation? I’d be glad to talk about the problem.

Read my résuméDownload PDFLinkedIn (opens a new tab)