Akankhya Singh
Data Analyst & AI Strategist
MSc Business Analytics (UCD Smurfit) | Specializing in Market Intelligence & Autonomous Workflows.
Hi, I'm Akankhya Singh.
I am a Data Analyst & AI Strategist dedicated to moving businesses beyond traditional reporting into the era of Agentic AI. With a technical foundation in Computer Science and a Master's from UCD Smurfit, I specialize in building data systems that don't just "show" insights but "act" on them.
My Focus for the UAE Market: As Dubai accelerates toward its D33 Economic goals, I provide the bridge between Big Data and Autonomous Action. My work centers on:
- •Precision Analytics: Identifying six-figure AED growth opportunities — such as my Dubai-Cashless 2026 initiative, which projected AED 250K+ in annual recoverable savings through optimized logistics and checkout-to-fulfillment analysis.
- •Agentic Orchestration: Developing autonomous agents (using LangChain and Python) that automate complex business logic and personalized customer engagement.
- •Cloud-Native Insights: Architecting robust pipelines on Azure and BigQuery (GCP) to ensure data integrity at scale.
Relocating to Dubai, I am looking to partner with tech-forward organizations ready to pioneer the next generation of intelligent, automated decision-making.
Skills
Skills
LangChain, CrewAI, Agentic RAG.
Context
Building autonomous agents that don't just predict trends but trigger business actions—like my Retention Pulse system that isolated AED 1.52M in revenue at risk across 351 high-risk VIP accounts.
Skills
Python (Pandas/Scikit-learn), Advanced SQL, ETL Pipelines, Azure, BigQuery (GCP).
Context
Architecting scalable, cloud-native pipelines capable of handling 1M+ monthly data points with 99.5% integrity, ensuring a robust foundation for AI deployment.
Skills
Power BI (Advanced DAX), Market Quantification, ROI Analysis.
Context
Translating complex datasets into executive financial impact—such as identifying AED 251K+ in annual logistics leakage to support Dubai's D33 economic goals.
Certifications
My professional credentials and qualifications.
Google / Coursera
Bloomberg LP
Dubai Future Foundation
dbt Labs
Work Experience
My professional journey.
Oct 2025 – Jan 2026
- Supported the strategic roadmap for a UK-based client's market expansion, combining market opportunity quantification with workflow automation to enable a high-impact, low-friction growth strategy.
- Enabled 25+ consultants with live analytics for client strategy and planning, improving consulting efficiency and enterprise performance management.
- Architected a market diversification approach that delivered 160% ROI and automated 36% of manual end-to-end workflows.
- Quantified efficiency gains and risk exposure to identify £3,200 in annual savings per boutique firm.
- Reduced weekly data preparation time by 15% through workflow optimization, improving turnaround for time-sensitive client deliverables.
Jun 2022 – Mar 2024
- Built and maintained the data integrity foundation for enterprise banking systems, ensuring reliability at scale.
- Maintained 100% uptime of a live banking website, preventing transaction failures across 1M+ monthly data points.
- Led governance initiatives achieving 99.5% improvement in data integrity, strengthening FP&A accuracy.
- Delivered customized reporting and market intelligence to identify risks, opportunities, and performance gaps for senior stakeholders.
- Translated complex datasets into actionable insights for business performance analytics and strategic planning support.
Projects
My work in Agentic AI and data systems.
Retention Pulse Agent
Revenue at Risk
AED 1.52M
VIPs at High Risk
351
Avg. Support Tickets (At-Risk Users)
5.98
Executive Summary
In high-volume e-commerce, customer churn is a silent revenue killer. Traditional analytics report on churn after it happens. The Retention Pulse Agent takes a statistically rigorous, proactive approach instead of a reactive one.
- •Engineered modular dimensional data models (dim_customers, fct_churn_signals) in dbt Core, with automated data quality tests across transactional data, isolating AED 1.52M in revenue tied to 351 high-risk VIP accounts.
- •Designed a statistical A/B testing framework evaluating retention interventions against a holdout control group, measuring statistically significant churn reduction (p < 0.05) before rolling any intervention out.
- •Orchestrated a stateful LangGraph agent with deterministic tool-calling to run SQL warehouse queries directly and auto-generate executive decision memos — turning a validated test result into a same-week business recommendation.
// Technical Logic Flow
Data Ingestion (SQL) & dbt Core Modeling
└──> Dimensional Models (dim_customers, fct_churn_signals) + Automated Data Quality Tests
└──> Risk Scoring & Segmentation (351 High-Risk VIPs Identified)
└──> A/B Test Design: Holdout Control Group
└──> Statistical Significance Testing (p < 0.05)
└──> Agentic Workflow (LangGraph)
├── Execute SQL Warehouse Queries
├── Synthesize Executive Decision Memo
└── Deliver Validated, Actionable RecommendationDashboard Preview
A Power BI dashboard for the Retention Pulse VIP Rescue project.
Explore the full data model and interactive visuals by downloading the PBIX file.
Project Technology Stack
Dubai Cashless 2026: Reducing Logistics Leakage
Annual Logistics Leakage
AED 251,350
Average RTO Rate
17.43%
Model Recall
88%+
Executive Summary
In direct alignment with the Dubai Economic Agenda (D33), this project optimizes e-commerce logistics by identifying financial leakage in Cash-on-Delivery (COD) cycles. By analyzing localized delivery data across Dubai districts, the system identifies AED 251,350 in annual operational waste.
- •Architected an end-to-end predictive classification pipeline (XGBoost) to identify high-risk COD orders prone to Return-to-Origin (RTO) failure, achieving 88%+ recall.
- •Designed SQL and dbt transformation workflows tracking checkout velocity, address validity, and historical fulfillment to isolate the highest-risk segment driving the 17.43% RTO rate.
- •Built an automated intervention framework reducing delivery overhead by 12.5%, triggering an Agentic AI Nudge that proactively encourages customers toward Aani Instant Payments to secure high-risk deliveries.
// Technical Logic Flow
Data Ingestion (SQL): Aggregating transaction logs, district codes, and payment methods.
└──> Feature Engineering (Python): Analyzing payment method risk profiles — COD shows significantly higher return rates than digital alternatives.
└──> RTO Prediction (XGBoost): Classifying orders as 'Safe' or 'High-Risk', achieving 88%+ recall.
└──> Risk Scoring & Segmentation: Isolating the 17.43% of orders driving the majority of logistics waste.
└──> Agentic Nudge (LangChain):
├── Analyze high-risk delivery context
├── Generate personalized Aani payment links
└── Trigger automated "Digital-First" outreachDashboard Preview
A Power BI dashboard for the Dubai Cashless 2026 project.
Explore the full data model and interactive visuals by downloading the PBIX file.