About Hasan
🔎 Data Engineer & Applied Data Scientist | End-to-End Data Platforms — from Ingestion to ML to Dashboard
Most data engineers can't take a model into production. Most ML specialists can't build the platform underneath it. I do both — one freelancer who can design, build, and ship the complete system.
🔧 What I build
• Production data platforms on AWS (S3, ECS Fargate, RDS, EventBridge)
• ELT pipelines & warehouse modeling with dbt + PostgreSQL
• Orchestration with Apache Airflow
• ML that ships: XGBoost / scikit-learn models served via FastAPI — not left in notebooks
• Streamlit + Plotly dashboards
📊 Recent work
⚡ Zephyrwerk Energy Analytics — live German electricity market + weather data on AWS: dbt ELT with 65k+ tested hourly records, Parquet lake on S3, XGBoost day-ahead price forecasting via FastAPI, daily EventBridge scheduling, full CI.
🚆 Mobility Delay Platform — Deutsche Bahn delay analytics & prediction: Airflow ingestion, dbt models with schema tests, Random Forest /predict API, 36 automated tests, fully Dockerized.
🛒 E-commerce analytics — root-caused a 28% AOV decline in an 85k-row transaction dataset and turned a retention gap into recommendations worth ~€1.25M.
🏭 Before freelancing: 4 years in industrial AI/ML — real-time computer vision at 10+ parts/sec, 90%+ defect detection, ~15–20% production efficiency gains for manufacturing clients.
🤝 How I work
• Your business problem first, then the tech stack
• Proactive communication — no black boxes
• Tested, modular code + clean handover: README, architecture docs, deployment guide
📍 Based in Munich, available remotely. English (C1), German (B2).
📩 Need a platform built, a pipeline made production-grade, or a model taken from notebook to API? Send me a message.
German
Conversational
English
Fluent
Turkish
Native or bilingual
Can work on-site
Munich (up to 30km), Nuremberg (up to 30km)
Experience
- Zephyrwerk EnergyZephyrwerk Energy Analytics PlatformENERGY AND UTILITIESJune 2026 - Today (1 month)Munich, GermanyProduction-grade data platform ingesting live German wholesale electricity market data (SMARD) and weather data (Open-Meteo) to support renewable energy trading and operations decisions. Built local-first with LocalStack, deployed to AWS with no code changes between environments — only env vars differ.
- 3-layer ELT architecture (raw → staging → analytics) on PostgreSQL with dbt Core: 4 staging models, 5 fact tables, and a date dimension covering 65k+ hourly observations from 2019 to present, fully tested and documented
- Idempotent ingestion pipeline writing partitioned Parquet to S3 (Hive-style year=YYYY/month=MM), serving both historical backfill and daily incremental modes from a single code path
- FastAPI REST API following dependency injection and repository patterns, serving analytics and ML predictions with auto-generated OpenAPI docs
- XGBoost forecasting models for day-ahead electricity price and renewable generation, serialized to S3 and loaded by the API on startup
- Multipage Streamlit dashboard with Plotly charts consuming the FastAPI service
- pytest unit and integration tests using moto and unittest.mock for AWS and HTTP isolation, automated via GitHub Actions CI (ruff + pytest) with branch-protected main
- Deployed to AWS ECS Fargate with EventBridge daily scheduling, S3 as model registry, and RDS PostgreSQL as the analytics store
GitHub: - German E-Commerce CompanyData AnalystE-COMMERCEApril 2026 - April 2026Munich, GermanyMarketing Analytics | Python (pandas, Matplotlib, seaborn) | Transaction Data
- Cleaned and prepared a raw 85,000-row transaction dataset through deduplication, invalid order filtering (16.6% of records), and euro-cent format correction across three mixed formats, producing a reliable 64,555-row analytical base
- Defined a three-KPI framework (Average Order Value, Repeat Purchase Rate, Monthly Active Customers) to assess revenue quality, customer loyalty, and acquisition efficiency independently and in combination
- Identified a sustained 28% AOV decline (€273 → €197) through four-hypothesis root cause analysis — systematically eliminating shipping artifacts, coupon effects, and country mix shift before confirming H2 product mix shift toward budget items (€0–30) as the primary driver
- Uncovered a structural retention gap: 78.1% of customers purchased only once, yet repeat customers generated 3.7× more annual revenue (€754 vs. €203), making retention the highest-ROI lever available
- Revealed acquisition-led growth pattern: new customers represented 73–88% of monthly actives throughout the year, with 110% H2 customer growth driven by two identifiable campaign pulses
- Translated findings into three concrete recommendations — free shipping threshold, post-purchase retention programme, and lost customer reactivation — with a combined conservative revenue opportunity of ~€1.25M
- Personal ProjectMobility Delay Prediction PlatformFebruary 2026 - May 2026 (3 months)Munich, GermanyDesigned and built a production-style data engineering platform that ingests live railway timetable data from the Deutsche Bahn API and transforms it into ML-ready datasets for delay analysis and prediction.The system implements a layered architecture (raw → staging → analytics) with modular ETL pipelines, automated delay calculations, and feature engineering — all running in a Dockerized PostgreSQL environment with idempotent ingestion and JSONB storage for full reproducibility.Key technologies: Python · PostgreSQL · FastAPI · Docker · XML parsing · SQLAlchemy
Recommendations
Be the first to recommend Hasan
Help this freelancer shine by sharing your experience working together.
These freelancer profiles also match your criteria
Agatha Frydrych
Backend Java Software Engineer
4.7
(3)
2
Baptiste Duhen
Fullstack developer
4.6
(4)
5
Amed Hamou
Senior Lead Developer
4
(2)
7
Audrey Champion
Web developer
4.3
(3)
4
Education
- Bachelor of Engineering in Electronics and Communication EngineeringIstanbul Technique University2020Bachelor Electronics and Communication Engineering
- Master of Science in Computer EngineeringIstanbul Technique University2024Master Computer Engineering
Certifications
- Data ScienceWBS Coding School2025