AI Transformation · BFSI · E-commerce · Telecom

Swati KeshariAI Product Analyst

Bridging business strategy and AI delivery.
I find the broken process · design the automation · ship the product · govern the outcome.

I operate at the intersection of
AI Business Analysis AI Product Management Automation Strategy
8+
Years
5
RPA Bots
12+
Depts Automated
35%
Resolution ↑
30%
Effort ↓
5
Live AI Apps
AI Use Case IdentificationLLM IntegrationAgentic AI SystemsBRD · FRD · RTMPrompt EngineeringRPA · IPA AutomationAML & Fraud DetectionAI GovernanceProduct RoadmapRAG ArchitectureStakeholder ManagementData MigrationBPMN Process MappingAgile DeliveryAI Ethics & CompliancePower BI DashboardsModel Acceptance CriteriaSQL Analytics AI Use Case IdentificationLLM IntegrationAgentic AI SystemsBRD · FRD · RTMPrompt EngineeringRPA · IPA AutomationAML & Fraud DetectionAI GovernanceProduct RoadmapRAG ArchitectureStakeholder ManagementData MigrationBPMN Process MappingAgile DeliveryAI Ethics & CompliancePower BI DashboardsModel Acceptance CriteriaSQL Analytics
Core Expertise

What I Bring to Every Engagement

Eight years of BA rigour and AI product thinking across BFSI, e-commerce, and telecom — applied to the most complex automation and AI programmes in the market.

Business Analysis
AI Use Case IdentificationBRD / FRD DocumentationRequirements ElicitationProcess Mapping (BPMN)Stakeholder ManagementUAT Planning & ExecutionAgile / Scrum DeliveryData Analysis & SQLAI Governance FrameworksModel Acceptance CriteriaCurrent vs Future State DesignBacklog ManagementUser Stories & Acceptance CriteriaRTM & TraceabilityCross-functional FacilitationSprint Planning & Grooming
Product Management
Product Roadmap StrategyAI Product LifecycleLLM Product DesignAgentic AI SystemsExecutive StorytellingKPI & Metric DefinitionScope Trade-off ManagementProduct-Market FitGo-to-Market PlanningROI & Business Case DevelopmentWireframing (Figma / Balsamiq)Version-Controlled Product DocsAPI Integration StrategyAdoption & Change Management
GenAI & Technical Fluency
Prompt EngineeringRAG ArchitectureLLM IntegrationAgentic AI OrchestrationAI Ethics & FairnessModel Drift MonitoringHallucination ControlsVector DatabasesAnthropic APIAI Ethics (GDPR · DPDP · EU AI Act)MLOps AwarenessDocument AI · OCR
Domain Expertise — BFSI · Fintech · E-commerce
AML / KYC ComplianceFraud Detection SystemsCredit Risk AutomationPayments & SettlementRegulatory Technology (RegTech)FCA / RBI ComplianceVendor ManagementDigital TransformationData GovernanceAudit Readiness
🎯
Business Analysis
The foundation that makes AI projects succeed — eight years of rigour across the most demanding regulated environments.
AI Use Case Identification
BRD / FRD / Requirements
Process Mapping & BPMN
RPA / IPA Automation Design
Stakeholder Management
Agile Delivery & UAT
AML / Fraud / Compliance
🚀
Product Management
Product strategy, roadmap ownership, and AI solution design — from first principles to shipped product.
Product Roadmap & Strategy
AI Product Lifecycle
Prompt Engineering
Wireframing (Figma/Balsamiq)
Backlog & Prioritisation
KPI & Metric Definition
Version-Controlled Product Docs
📊
Data & Analytics
SQL root cause analysis, Power BI dashboards, source-to-target mapping, data quality validation. Santander AML: 28% fewer false positives via SQL-driven model tuning. Amazon: real-time vendor SLA monitoring replacing 6 manual Excel trackers.
SQLPower BIData Governance
🔐
Compliance & RegTech
AML workflow redesign, fraud dispute management, regulatory UAT, InfoSec sign-off. FCA, GDPR, RBI, and SOX-aware delivery. BFSI hiring data 2026: AML/KYC compliance is the #1 driver of fintech hiring — my core domain.
AMLGDPRFCA · RBI
🌐
Digital Transformation
Enterprise-wide digital adoption, paper-to-digital migration, app modernisation (Flutter), and cross-functional change management. BT app: 40% performance boost. Amazon: eliminated paper-based workflows across 12 departments.
FlutterChange MgmtAdoption
Career

Where I've Delivered

Tier-1 global programmes. Every role centred on translating complex business problems into scalable, measurable solutions.

MAR 2024 — PRESENT
NTT DATA
Lead Business Analyst · Gurugram
Leading simultaneous BA workstreams across Citi, Santander UK, and British Telecom. Delivering IPA automation, data migration governance, AML redesign, and Flutter app modernisation across global stakeholder groups — maintaining delivery velocity across three live programmes.
Citi IPA — 100+ tasks/week eliminated Citi CAI Data Migration Santander AML — 28% false positive reduction BT Flutter — 40% performance boost
AUG 2022 — FEB 2024
Amazon
Consultant BA · EU Region · Bengaluru
Led AI-augmented automation discovery and delivery across 12+ EU vendor management departments. Designed prioritisation frameworks, automated workflows, and drove enterprise-wide digital adoption. Coordinated InfoSec and Legal for compliance-led rollouts across jurisdictions.
30% manual effort reduction 20% SLA compliance improvement 12+ departments automated
DEC 2019 — AUG 2022
Barclays
Senior Business Analyst · Noida
Redesigned fraud and dispute workflows for Barclaycard using AI-assisted classification and rules automation. Built SOPs, BPMN process maps, and Power BI dashboards. Led regulatory UAT for FCA compliance features in collaboration with product and technology teams.
35% faster dispute resolution Power BI dashboards Regulatory UAT lead
JAN 2019 — DEC 2019
ICICI Bank
Deputy Manager — BA · Noida
Automated verification workflows improving documentation accuracy by 40%. Contributed to CRM module rollouts to streamline reporting and client management across retail banking operations.
40% documentation accuracy ↑ CRM rollout
Signature Projects

Automation Delivered End-to-End

Two programmes that define my approach — from discovery to production, with measurable outcomes and lessons that shaped how I work.

Citi × THD CanadaGreenfield RPA
ARTP — Automated Routing to Processing
Extended Line Accommodation Program from USA to Canada. Zero prior automation. Sole BA — designed 5-bot architecture from discovery through production.
1
AS-IS Discovery — Documented 5 fully manual workflows. No audit trail, no scheduling, high FTE cost. Gap analysis against USA automation model identified all candidates.
2
5-Bot Architecture Design — Each bot: single responsibility, defined trigger, precise I/O specification, error handling and alert logic for every exception scenario.
3
Full BRD Suite — Process flows, input-output schemas, encryption requirements, fallback processes, UAT cases, impact checklist for go-live sign-off.
4
Production Lesson — Loader Bot silent failure due to missing fallback for malformed emails. Now: fallback processes and alert triggers are mandatory in every specification I write.
Outcomes
5
Bots Designed
100%
Manual Eliminated
Full Audit Trail
Amazon EUCross-Functional
Vendor Management Automation
12+ departments. Manual Excel-based tracking transformed into automated performance monitoring, real-time SLA alerting, and compliance reporting.
1
Discovery Across 12 Departments — 4 weeks of workshops and walkthroughs. 30 pain points → 8 automation candidates → 3 quick wins via Feasibility × Impact scoring framework.
2
Core Process Redesign — Weekly vendor review: 6 Excel files → 3-day manual cycle → 8 handoffs → zero audit trail. Redesigned to automated ingestion, real-time alerts, exception routing.
3
Compliance-Led Delivery — Data sources, SLA threshold logic, alert triggers, InfoSec and Legal sign-off coordinated across all 12 departments and EU jurisdictions.
4
Production Lesson — No monitoring plan defined at deployment. Drift went undetected for weeks. Now: post-deployment monitoring plan is a mandatory deliverable in every programme.
Outcomes
30%
Effort Saved
20%
SLA Improvement
12+
Depts Done
ARTP Bot Architecture

5-Bot Assembly Line — Designed End-to-End as Sole BA

Single responsibility · Precise handoff specs · Scheduled triggers · Full audit trail

📨
Loader Bot
Email ingestion, account scraping, data extraction
⚙️
Processing Bot
Scheduled, accommodation rules, 2 structured outputs
📬
Response Bot
THD email taxonomy matching, master file updates
🔐
Encryption Bot
Validate, encrypt to spec, deliver, confirm
Decline Bot
End-to-end decline workflow, comms, audit trail
AI Depth

GenAI · Agentic AI · Prompt Engineering · RAG

What hiring managers are testing for in 2026 — and how I apply each concept in real delivery contexts.

Generative AI
GenAI for Business
GenAI handles unstructured content, generates outputs, and reasons through ambiguity. For BA/PM: AI requirements need model accuracy thresholds, hallucination controls, fallback processes, and explainability requirements — especially critical in FCA and RBI-regulated environments. Standard BRD templates miss all of this.
Applied: Built 4 production AI agents on Anthropic API with defined system prompts, JSON output schemas, and governance documentation.
Agentic AI
Multi-Agent Orchestration
Multiple AI models orchestrated into autonomous workflows — each agent with a defined role, passing structured outputs to the next. Architecturally identical to the 5-bot ARTP chain I designed as BA, but with AI reasoning at each step. New BA questions: handoff conditions, confidence thresholds, chain failure escalation, human override triggers.
Applied: Designed and shipped 4-agent AI toolkit with inter-agent handoffs, structured outputs, and Notion API integration.
Prompt Engineering
Requirements for AI Models
Prompt engineering is BA requirements applied to AI — precision of instruction determines quality of output. In practice: defining the AI's role, output schema, edge case handling, constraint logic, and tone. I write production system prompts the same way I write BRDs — with acceptance criteria, fallback conditions, and version control.
Applied: 4 production system prompts with JSON schemas, edge case handling, and versioned in Git alongside product decision docs.
Retrieval Augmented Generation
RAG — What Every AI BA Must Know
LLM responses grounded in documents retrieved at query time. Critical for enterprise AI where answers must reference proprietary policies, contracts, or regulated content. My 7-phase BA checklist covers everything from indexing strategy to post-deployment monitoring.
INDEXING
Knowledge base scope, chunking strategy, metadata tagging, PII controls, re-indexing cadence
RETRIEVAL
Accuracy thresholds, top-K params, relevance scoring, fallback when no match found
GENERATION
Hallucination rate criteria, prompt template ownership, citation requirements
GOVERNANCE
Audit trail, human review gates, GDPR · DPDP · EU AI Act compliance
Modern AI PM Practice
Product Decisions in Git
The modern AI PM lives in the same tools as engineers — version-controlled docs, markdown files in the repo, PRs with product reasoning attached. When an engineer asks "why did we build it this way?" — git log shows them.
What I push alongside code
PRD.md — Product requirements, versioned
DECISIONS.md — Why we built it this way
PROMPTS.md — System prompts with version history
METRICS.md — Success criteria & monitoring plan
ROADMAP.md — Living strategy, diff-able over time
Tools & Technologies
JiraConfluenceSQLPower BILucidchartBalsamiqFigmaPostmanBPMNGoogle Data Studio Anthropic APIPrompt EngineeringClaude AINotion API GitHubEmergentLovableReplit RPA / IPAAAC (IIBA)PSPO (Trained)Six Sigma
Built & Deployed

Live AI Products & Agents

Six deployed products — from production AI agents to client-facing web apps. Each one encoding BA and PM methodology into working software that anyone can use right now.

AI BA Toolkit — 4 Agents
LIVE
Production AI toolkit — use case scanner, AI-ready BRD generator, process blueprint, governance framework. Anthropic API + Notion integration. Every section built from real delivery experience across Citi, Amazon, and Barclays.
Embedded in this page ↑
💡
AI Use Case Assistant
LOVABLE
Describe any business context — pain points, goals, KPIs, constraints — and the app generates a full AI transformation strategy. Operationalises the discovery framework I ran manually at Amazon across 12 departments.
Open App →
🧠
AI Consulting Mastery
REPLIT
Interactive AI consulting dashboard — frameworks, methodologies, and strategic tools for AI-led business transformation. Designed to demonstrate AI BA and consulting capability across real enterprise scenarios.
Open App →
📦
VendorIQ
EMERGENT
AI Vendor Intelligence Platform — centralised profiles, performance scoring, risk flags, automated SLA alerts. The product Amazon EU's vendor management needed. Built from the exact requirements I designed at Amazon.
Open App →
🔐
AI KYC Onboarding
EMERGENT
AI-powered KYC — document validation, compliance routing, automated decisions with human override gates. Built from Barclays fraud and Santander AML experience. Mirrors the compliance logic from the ARTP Decline Bot.
Open App →
📋
RAG Requirements Checklist
DOCUMENT
7-phase BA checklist for RAG projects — knowledge base scoping, retrieval accuracy thresholds, hallucination controls, PII governance, audit trail, and monitoring plan. Fully readable in this portfolio.
Education & Certifications
IIBA · Certified
AAC
Agile Analysis Certified
IIBA · In Progress
CBAP
Certified BA Professional
Scrum.org · Trained
PSM1 · PSPO
Scrum Master & Product Owner
Sharda University
MBA
Banking & Financial Services
Let's Build Something
Ready to Deliver AI Transformation

Open to AI BA · AI PM · Lead BA · Automation Strategist roles
Noida · Lucknow · Pune · Hyderabad · Bengaluru · Remote