Mohamed Mosaed — Senior Data Analyst | Data Scientist

Career Impact

Business outcomes backed by technical depth.

A quick view of the scale, outcomes, and decision systems delivered across enterprise analytics environments.

50M+records processed across financial, claims, clinical, and operational datasets

200+business users supported through reporting modernization and KPI platforms

60%report generation / data quality improvement across key initiatives

30%ETL runtime reduction through PySpark and Databricks optimization

25%manual review reduction by improving underwriting analytics workflows

7+years across insurance, healthcare, consulting, and financial services

What I Build

From raw data to production-ready analytics products.

AI-Powered Analytics

RAG tools, semantic search, OpenAI API workflows, NLP extraction, and AI-assisted knowledge retrieval for business teams.

Cloud Data Pipelines

PySpark and Databricks pipelines connected to Azure SQL, Azure Data Lake, SQL Server, and downstream BI layers.

Predictive Modeling

Classification, regression, survival analysis, anomaly detection, feature engineering, validation, and performance tuning.

Executive BI Products

Power BI, Tableau, R Shiny, DAX, Python visuals, KPI governance, and dashboards designed for decision-making.

Detailed Featured Projects

Case studies with tools, usage, architecture, and impact.

Each project now explains not only what was built, but exactly what tools were used and how they contributed to the solution.

Insurance · AXA XL

Renewal Radar — AI-Powered Policy Renewal Intelligence Platform

A Generative AI and analytics platform that helped underwriters identify expiring policies, retrieve account intelligence, and reduce time spent searching annual reports and business documents.

Gen AI / RAG

Business Problem

Underwriters spent up to 2 weeks manually collecting financial information, business descriptions, contacts, and renewal context for accounts approaching expiration. The manual workflow delayed renewal action and increased the risk of missed opportunities.

Technical Approach

Created a 90-day renewal pipeline to identify upcoming policy expirations.
Built a document intelligence workflow for 500+ annual reports and business documents.
Used semantic search and OpenAI responses to return concise account-level insights.
Delivered insights in Power BI so underwriters could act inside their existing workflow.

Tools Used & How I Used Them

PySparkProcessed high-volume policy, account, and claims records; joined renewal dates, account metadata, premium history, and underwriting attributes at scale.

DatabricksHosted the distributed pipeline, scheduled transformations, managed notebooks, and created reliable curated tables for downstream reporting.

PythonHandled document parsing, text cleaning, chunking logic, API orchestration, and automation around document intelligence workflows.

OpenAI APIGenerated natural-language summaries, extracted business context, and allowed underwriters to ask questions about account documents.

Vector EmbeddingsConverted annual report chunks into searchable semantic representations so relevant text could be retrieved even without exact keyword matches.

Power BIBuilt underwriter-facing dashboards showing expiring accounts, renewal priority, extracted insights, and action-ready account views.

Policy Data→PySpark Renewal Pipeline→Document Chunking→Embeddings + Semantic Search→OpenAI Summary→Power BI

2 weeksmanual research cycle eliminated

23%renewal efficiency improvement

<30 sectime to retrieve critical account insights

PySparkDatabricksPythonOpenAI APIRAGVector EmbeddingsPower BIAzureSQL

Insurance · AXA XL

DataGuard — Deep Learning Anomaly Detection & Data Quality Engine

A proactive data quality monitoring system that detects unusual policy records and underwriting exceptions before flawed data reaches downstream reporting or modeling layers.

Deep Learning

Business Problem

Bad data was often discovered only after it moved into reporting, dashboards, or predictive models. This created manual cleanup, delayed analysis, and lowered trust in analytics outputs.

Technical Approach

Created validation checks for incoming policy and transaction records.
Used embeddings to represent text-heavy policy attributes and categorical fields.
Trained autoencoder models to learn normal data patterns and flag outliers.
Added alerting and exception reporting so analysts could investigate issues early.

Tools Used & How I Used Them

BERT EmbeddingsConverted policy text, descriptions, and string-heavy fields into dense numerical features for anomaly detection.

AutoencodersLearned normal record patterns and produced reconstruction-error scores to detect unusual or potentially incorrect records.

TensorFlow / PyTorchUsed for model experimentation, neural network training, threshold testing, and anomaly scoring workflows.

PythonHandled feature engineering, preprocessing, model evaluation, scoring logic, and integration with the validation layer.

PySparkScaled scoring and data preparation across large underwriting datasets before feeding records into the detection workflow.

Power BIDisplayed anomaly trends, exception counts, policy-level details, and quality monitoring KPIs for business review.

Incoming Records→Feature Engineering→BERT Embeddings→Autoencoder Scoring→Risk Flags→Quality Dashboard

60%data quality improvement

Proactiveshift from cleanup to prevention

Lower riskbad data caught before downstream reporting

BERTAutoencodersTensorFlowPyTorchPythonPySparkAnomaly DetectionPower BI

Insurance · AXA XL

UnifiedView — Enterprise Dashboard Consolidation Platform

A reporting modernization effort that consolidated fragmented dashboards and data sources into one governed Power BI experience for underwriting and leadership teams.

BI Engineering

Business Problem

More than 200 business users relied on 10 separate dashboards and files across SQL Server, Excel, and Databricks. Metrics were inconsistent, reporting was slow, and teams lacked a single source of truth.

Technical Approach

Mapped existing dashboards, metric definitions, refresh schedules, and source systems.
Moved key datasets into curated Databricks tables and SQL views.
Rebuilt KPI definitions using a governed DAX metric layer.
Designed one executive dashboard with drill-downs for underwriting teams.

Tools Used & How I Used Them

Power BIDesigned the final dashboard experience, drill-through pages, slicers, KPI cards, executive views, and underwriting performance pages.

DAXRebuilt standardized measures for loss ratios, policy counts, premium metrics, trend calculations, and KPI comparisons.

DatabricksCentralized source data and created clean curated tables to support the Power BI semantic model.

SQL ServerExtracted legacy reporting data, validated source logic, and created migration checks between old and new reporting layers.

Python VisualsBuilt custom Matplotlib and Plotly visuals for analysis that Power BI could not easily render natively.

Data GovernanceDocumented KPI definitions, refresh logic, data owners, and validation rules to reduce conflicting reports.

SQL / Excel / Databricks→Curated Data Model→DAX KPI Layer→Power BI Semantic Model→Executive Dashboard

200+business users supported

28%analyst productivity improvement

60%faster report generation

Power BIDAXSQL ServerDatabricksPythonPlotlyMatplotlibData Governance

Clinical Research · UMass Chan

ClinicalCore — SQL Data Architecture Redesign for Research Infrastructure

A clinical data architecture cleanup that transformed hundreds of undocumented tables into governed, analysis-ready data views for researchers and clinical analytics teams.

Data Engineering

Business Problem

Research teams worked across 700+ undocumented tables with inconsistent naming, duplicated fields, unclear refresh logic, and limited documentation. Analysts wasted hours finding the right data before analysis could begin.

Technical Approach

Audited table usage, schema quality, field definitions, and redundancy.
Created clean SQL views for frequently used research datasets.
Documented source tables, refresh cadence, business definitions, and use cases.
Supported dashboards and analysis for multiple active clinical studies.

Tools Used & How I Used Them

SQL ServerAudited schemas, wrote optimized SQL views, joined patient and study-level tables, and simplified access to trusted datasets.

Python / PandasProfiled data quality, checked missingness, validated distributions, and prepared analysis-ready outputs for research teams.

Power BICreated monitoring dashboards for patient cohorts, recruitment, enrollment, and clinical study progress.

Data DictionaryDocumented field meanings, table relationships, source logic, and recommended usage for analysts and researchers.

Governance ProcessDefined naming conventions, retired unused tables, and improved consistency across clinical reporting assets.

Statistical AnalysisSupported logistic regression, survival analysis, population risk analysis, and publication-ready summary tables.

700+ Tables→Schema Audit→Clean SQL Views→Documentation→Dashboards + Studies

70%faster data access

500K+patient records supported

1,000+patients tracked across studies

SQL ServerPythonPandasPower BIClinical DataGovernanceSurvival Analysis

Insurance · KPMG

MarketCompass — Geospatial Insurance Market Expansion Analytics Tool

A geospatial analytics and predictive modeling solution that helped a life insurance client compare U.S. states and identify strong regional expansion opportunities.

Predictive ML

Business Problem

A life insurance client needed a data-driven way to compare states for regional expansion. Leadership needed to evaluate demographics, market opportunity, risk, and growth indicators in a single framework.

Technical Approach

Collected and transformed market, demographic, and insurance-related indicators.
Built a scoring framework across client-defined expansion criteria.
Created a state comparison dashboard with map-based exploration.
Added ML recommendation logic to rank markets and explain key drivers.

Tools Used & How I Used Them

Power BI MapsBuilt geospatial dashboards showing state-level scores, filters, comparison views, and opportunity segmentation.

Scikit-learnBuilt predictive and ranking models to score potential markets and recommend high-opportunity states.

PythonCleaned market datasets, engineered features, normalized variables, and prepared modeling tables.

PlotlyCreated interactive charts for state comparisons, factor breakdowns, and executive storytelling visuals.

SQLJoined and structured data sources into repeatable tables used by the dashboard and modeling layer.

Consulting FrameworkTranslated business criteria into measurable scoring dimensions that leadership could understand and defend.

Market Data→Feature Engineering→State Scoring Model→ML Recommendation→Power BI Map

Strategicmarket expansion asset

Data-backedstate comparison framework

Executivedecision-ready dashboard

Power BIPythonScikit-learnPlotlyGeospatial AnalyticsSQLMarket Scoring

Insurance · KPMG

DocuMind — NLP Contract Digitization & Cloud Migration Engine

An NLP and OCR pipeline that converted physical and PDF contracts into structured records, accelerating a client cloud migration and reducing manual data extraction effort.

NLP / OCR

Business Problem

Thousands of physical and PDF contracts blocked cloud migration because key contract information existed only in unstructured documents. Manual entry was slow, expensive, and error-prone.

Technical Approach

Built OCR workflows to convert scanned contracts into machine-readable text.
Used NLP preprocessing to clean and structure extracted text.
Applied classification models to detect key fields and contract categories.
Loaded structured contract outputs into database tables for migration.

Tools Used & How I Used Them

OCRExtracted text from scanned contracts and PDF documents to create a machine-readable document layer.

SpaCyUsed named entity recognition and linguistic preprocessing to identify parties, dates, clauses, and coverage terms.

NLTKPerformed tokenization, lemmatization, stemming, and text normalization for downstream NLP modeling.

Scikit-learn / XGBoostClassified extracted fields and document sections into structured categories for database loading.

PythonBuilt the end-to-end extraction pipeline, text cleaning functions, validation scripts, and export process.

SQLStored structured outputs, created QA checks, and prepared migration-ready contract tables.

PDF / Scanned Contracts→OCR→NLP Cleaning→Entity + Field Extraction→SQL Migration Tables

5 monthsmigration acceleration

30%accuracy improvement vs manual extraction

Structuredcontract data ready for cloud migration

OCRSpaCyNLTKPythonScikit-learnXGBoostSQLNLP

Insurance · AXA XL

AutoReport — Scheduled Excel Reporting Automation via VBA & Python

An automation workflow that replaced manual Excel refresh cycles with a scheduled VBA and Python process connected to Databricks SQL.

Automation

Business Problem

Analysts spent 15+ hours per week refreshing recurring Excel reports, copying data from Databricks, formatting outputs, and fixing human errors.

Technical Approach

Used VBA to trigger scheduled report refreshes from Excel.
Used Python to query Databricks SQL and apply transformations.
Automated formatting and output writing back into Excel templates.
Added error handling and notification logic for failed refreshes.

Tools Used & How I Used Them

VBACreated Excel macros to trigger the automation, refresh report tabs, and control workbook-level formatting.

PythonConnected to Databricks SQL, executed parameterized queries, transformed data, and wrote results into Excel templates.

PandasCleaned report outputs, performed calculations, reshaped data, and prepared final datasets for Excel delivery.

Databricks SQLServed as the source for scheduled reporting data and enabled repeatable extraction logic.

ExcelDelivered final reports in the business team's preferred format while removing manual refresh steps.

Error HandlingAdded validation checks and alerts so failed refreshes could be investigated quickly.

Excel Macro→Python Script→Databricks SQL→Pandas Transform→Formatted Report

15+ hrssaved per week

100%on-time report delivery

Zeromanual refresh errors

VBAPythonPandasDatabricks SQLExcelETL Automation

Architecture Thinking

How I think about systems.

These architecture patterns show the full path from source data to analytics delivery, not just final dashboards.

AI Retrieval Pattern

Unstructured documents are cleaned, chunked, embedded, retrieved with semantic search, summarized with OpenAI, and delivered through dashboards or analyst tools.

Data Quality Pattern

Incoming records move through validation, embeddings, anomaly scoring, exception rules, and quality dashboards before they reach downstream reporting.

BI Modernization Pattern

Fragmented sources are consolidated into curated tables, governed KPI layers, semantic models, and executive dashboards with clear ownership.

Technical Stack

Tools grouped by how I use them.

Data Engineering

PySparkDatabricksAzure Data LakeAzure SQLSQL ServerETL PipelinesData Modeling

Analytics & BI

Power BIDAXTableauR ShinyPlotlyMatplotlibExecutive Dashboards

AI & Machine Learning

OpenAI APIRAGBERTScikit-learnTensorFlowPyTorchXGBoostNLP

Experience Timeline

Cross-industry analytics background.

Sep 2023 – Present

Senior Data Analyst · AXA XL

Predictive modeling, underwriting analytics, PySpark/Databricks pipelines, RAG tools, anomaly detection, reporting modernization, and stakeholder-facing BI products.

Jun 2023 – Sep 2023

Data Analyst · UMass Chan Medical School

Clinical data architecture, patient-level dashboards, survival analysis, statistical modeling, SQL views, and publication-ready visualizations.

Jan 2022 – Jun 2023

Data Analyst · KPMG

Client-facing machine learning, NLP, dashboard automation, geospatial analytics, and consulting solutions for insurance clients.

Mar 2020 – Dec 2021

Data Analyst · McKesson

Healthcare data profiling, SQL/ETL development, predictive modeling, and operational analytics across clinical and supply chain datasets.

Let’s build data products that decision-makers actually use.

Open to Senior Data Analyst, Data Scientist, Analytics Engineer, and AI Analytics roles.

Email Call Resume