s.aSEFA BERK ACAR
FROM DATA TO DECISIONS

Sefa Berk Acar.

Data Scientist · Machine Learning · Applied AI

I turn complex data into machine learning solutions, analytical insights, and measurable business impact.

View My Work Contact Me Download CV
acarsefa621@gmail.com
BEYOND THE MODEL

Results that
make a difference.

+13%up to

Forecast accuracy improvementSelected SKUs

~5%

Unrealized sales identifiedOverall retail analysis

25%up to

SKU–store opportunity gapIndividual combinations

EXPLORE THE WORK BEHIND THE NUMBERS

01 / ABOUT

Mathematical thinking.
Practical solutions.

I’m a Data Scientist with a B.Sc. in Mathematics from Yildiz Technical University, applying machine learning and analytics to business problems.

My FMCG work spans forecasting, price elasticity, stock-out analysis, and unrealized demand. I also build customer analytics and Power BI reporting that supports decisions.

I’m exploring Quantum Computing and Quantum Machine Learning as emerging areas.

Grounded in mathematics

Statistics and optimization applied to complex problems.

Focused on business impact

Insights that inform pricing, inventory, and growth.

Curious about what’s next

Exploring quantum methods alongside applied data science.

02 / EXPERIENCE

Where data meets
the real world.

Applied work, measurable outcomes,
and knowledge shared along the way.

Reality Capital ManagementSeptember 2026 – Present

Data Scientist — Part-Time

Current
  • Analyze customer and business data to understand profiles, segments, locations, and behavior, and identify target customer groups.
  • Assess customer acquisition, retention, and business expansion opportunities through growth analysis.
  • Support market and competitor research focused on the Canadian market.
  • Create data-driven insights, reports, and visualizations that explain performance and growth opportunities and support marketing and strategy decisions.
Customer AnalyticsGrowth AnalyticsMarket AnalysisBusiness Intelligence
MiuulJul 2026 — Present

Quantum Machine Learning
Bootcamp Teaching Assistant

Current
  • Support teaching in Quantum Computing and Quantum Machine Learning.
  • Guide projects using Qiskit, quantum circuits, variational algorithms, and hybrid models.
Quantum MLQiskitHybrid Models
Logix AI LabsJan 2025 — Jul 2026Part-Time

Data Scientist

Applied impact
  • Built elasticity-aware forecasts for a leading FMCG client, improving accuracy by up to 13% on selected SKUs.
  • Estimated demand from SKU–store peer benchmarks, identifying ~5% unrealized sales overall and gaps of up to 25%.
  • Analyzed stock-out and on-shelf availability patterns to expose sales opportunities.
  • Built Power BI dashboards connecting sales and demand potential to pricing and inventory decisions.
PythonTime SeriesRetail AnalyticsPower BI
MiuulApr 2024 — Present

Data Science & Machine Learning
Bootcamp Teaching Assistant

Current
  • Support students in applied machine learning, statistics, time series forecasting, and model evaluation.
Machine LearningStatisticsTime Series Forecasting

03 / SELECTED WORK

Complex problems.
Meaningful outcomes.

From understanding demand
to informing the next decision.

Professional Work

Applied data science and machine learning work developed in real business environments.

DEMAND × PRICE
Time-series concept01 / 06

MACHINE LEARNING / RETAIL ANALYTICS

Price Elasticity &
Demand Forecasting

Problem
How do price changes affect FMCG demand?
Approach
Elasticity-aware features and time-series scenario modeling.
Impact
Up to 13% better accuracy on selected SKUs.
PythonPandasTime SeriesPower BI
Professional Case Study
FINDING THE UNSEEN
Peer demand
Actual sales
Potential beyond recorded sales
Peer-benchmark concept02 / 06

MACHINE LEARNING / BUSINESS ANALYTICS

Similarity-Based
Unrealized Demand Estimation

Problem
Estimate lost sales without reliable stock data.
Approach
SKU–store similarity benchmarks for counterfactual demand.
Impact
~5% unrealized sales overall; up to 25% per SKU–store.
PythonPandasSimilarity ModelingPower BI
Professional Case Study

Personal Projects

Open projects where code, methodology, and implementation can be explored directly.

FORECAST → PLAN → ACT
Historical salesInventory & shipments
  1. Forecast demandCompare forecasting models
  2. Simulate inventoryStock-outs · Service levels · Lost sales
  3. Recommend actionsAllocation · Transfers · Scenarios
From forecasts to inventory decisions03 / 06

DEMAND FORECASTING · INVENTORY OPTIMIZATION · DECISION SUPPORT

Demand Planning AI

An end-to-end Streamlit application connecting demand forecasts with operational inventory decisions.

Problem
Combine demand, stock, and planned shipments to identify risks and improve allocation.
Approach
Forecast demand, simulate stock, compare models, and test allocation scenarios.
Impact
Stock-out risk, service levels, lost-sales estimates, and transfer recommendations in one decision-support workflow.
PythonStreamlitTime Series ForecastingDemand PlanningInventory AnalyticsScenario Analysis
View on GitHub
04 / 06

PRODUCT ANALYTICS · EXPERIMENTATION

A/B Testing & Player
Retention — Cookie Cats

Problem
Does moving a progression gate affect player retention?
Approach
Hypothesis testing and bootstrapping.
Impact
Experimental evidence for product decisions.
PythonPandasStatisticsA/B Testing
View on GitHub
05 / 06

CUSTOMER ANALYTICS · CRM

CRM Analytics
— RFM & CLV Prediction

Problem
Understand customer behavior and value in coffee-shop transactions.
Approach
RFM, BG/NBD, and Gamma-Gamma models.
Impact
Customer segments and lifetime-value estimates.
PythonPandasLifetimesCRM Analytics
View Notebook
06 / 06

ARTIFICIAL INTELLIGENCE / OPTIMIZATION

AI-Based Secure
Route Optimization

Problem
Plan secure routes for critical missions.
Approach
AI optimization with terrain, risk, and drone inputs.
Impact
B.Sc. thesis on secure navigation.
PythonOptimizationGraph Theory
Academic Thesis

04 / COMPETITIONS & ACHIEVEMENTS

Curiosity, put
to the test.

Challenging datasets. Competitive settings.
Another way to keep learning.

5thPRIVATE LB

ING Hubs Türkiye Datathon

2025

Churn prediction with banking transaction and behavioral time-series data.

FinalistGini · Recall@10% · Lift@10%View Solution
8th90+ TEAMS

DatathonAI’24

2024

Image-based location prediction using ResNet50 embeddings and visual similarity.

FinalistL2 distance · Haversine distance · Spectral clusteringView Solution
TOP60.68 RMSLE

AlgoRun’24

2024

Retail sales forecasting using LightGBM and advanced feature engineering.

Top 6Time series forecasting · 0.68 RMSLE
11%TOP RANK

Trendyol E-Commerce Hackathon

2025

Search ranking and conversion prediction with learning-to-rank approaches and behavioral/session features.

32nd / 302 teamsMore than 1,100 entrants
MENTION PRIZE

Doping Technology Edu-Tech AI Hackathon

2024

Built an AWS-based educational analytics AI system, recognized for innovation and real-world impact.

Mention PrizeSelected among 450+ teams

05 / TOOLKIT

The tools behind
the thinking.

Data Science & Machine Learning

PythonScikit-learnLightGBMCatBoostTime Series ForecastingLearning to Rank

Data & Analytics

SQLPandasNumPyPower BIA/B Testing

Deep Learning & AI

TensorFlowPyTorchNLPReinforcement Learning

Deployment & Cloud

AWSFastAPIStreamlit

Mathematics

Linear AlgebraProbability & StatisticsOptimization

Quantum Computing

QiskitQuantum Machine LearningHybrid Quantum-Classical Models

06 / EDUCATION & CONTINUOUS LEARNING

A strong foundation.
An open mind.

GRADUATED FEBRUARY 2026

Yildiz Technical University

Bachelor of Science — Mathematics

BACHELOR’S THESIS

Determination of Optimum and Secure Routes for Mission-Critical Systems Using Artificial Intelligence Techniques

SupervisorAssoc. Prof. Dr. Muttalip Özavşar

Industry Thesis MentorAlper Bereketli, PhDSenior Principal Engineer, ASELSAN

Certifications

Learning is an ongoing practice.

Miuul

Data Scientist Bootcamp

European Coding School

Data Science and AI Specialization

Samsung Innovation Campus

Artificial Intelligence Program

DeepLearning.AI

Neural Networks and Deep Learning

DeepLearning.AI

Improving Deep Neural Networks: Hyperparameter Tuning, Regularization and Optimization

DeepLearning.AI

Structuring Machine Learning Projects

Akbank

Artificial Intelligence and Data Analytics 101

07 / WHAT’S NEXT?

Have a data problem
worth solving?

I’m open to Data Science, Machine Learning, Analytics, and Applied AI opportunities and collaborations.

acarsefa621@gmail.com
LinkedIn GitHub Kaggle