You will get a data warehouse with ETL pipelines and full history tracking
Project details
Most pipelines break the first time history changes. A customer moves city, the source record is overwritten, and every past order silently reports the new city. Nothing errors, and nobody notices until the numbers are questioned in a meeting.
I build warehouses that keep history intact. Every version of a changing record is preserved with its own valid_from and valid_to dates, so each transaction joins to the version that was true at the time and published reports stop moving. Loads are idempotent, so a failed run is recovered by simply running it again.
Five years in data engineering and analytics, including production Airflow pipelines into AWS and a BI migration onto a direct BigQuery connection. Everything ships as documented Python and SQL you can maintain yourself, with no proprietary tooling and no lock-in.
If a warehouse is overkill for your situation, I will tell you before you buy.
I build warehouses that keep history intact. Every version of a changing record is preserved with its own valid_from and valid_to dates, so each transaction joins to the version that was true at the time and published reports stop moving. Loads are idempotent, so a failed run is recovered by simply running it again.
Five years in data engineering and analytics, including production Airflow pipelines into AWS and a BI migration onto a direct BigQuery connection. Everything ships as documented Python and SQL you can maintain yourself, with no proprietary tooling and no lock-in.
If a warehouse is overkill for your situation, I will tell you before you buy.
Database Type
MySQL, MS SQL, Oracle, SQLite, PostgreSQL, MongoDB, Realm Database, Azure Cosmos DBWhat's included
| Service Tiers |
Starter
$95
|
Standard
$250
|
Advanced
$500
|
|---|---|---|---|
| Delivery Time | 3 days | 7 days | 14 days |
Number of Revisions | 1 | 2 | 3 |
Source Code |
About Muhammad
AWS, Data & AI Engineer | ETL Pipelines, LLM Evaluation | Python, SQL
Nuremberg, Germany - 8:22 pm local time
Five years across data engineering, ML and analytics. On the data side I have built production Airflow pipelines syncing third party APIs into AWS, migrated BI reporting onto a direct BigQuery connection to cut query costs, and rebuilt slow reporting SQL with window functions and CTEs so numbers were both faster and right.
On the AI side I architected agent training infrastructure in Python and LangChain that produced a 15% accuracy gain and a 12% tau improvement on benchmarks, led a team of 25 building training and regression datasets for rapid model versioning, and automated defect detection in training data, removing over 30 hours of manual review a week.
Data engineering:
• ETL and ELT pipelines from APIs, databases and files into your warehouse
• Dimensional modeling, including SCD Type 2 so historical reporting stops silently changing
• Idempotent loads that can be rerun safely after a failure
• SQL performance work on slow or incorrect reporting queries
• Orchestration with Airflow or cron
AI and LLM:
• Fine-tuning with SFT and RLHF
• Evaluation design: rubrics, benchmarks, and honest measurement of whether a change helped
• Training data generation and quality auditing
• RAG systems and agent tooling
Stack: Python, SQL, Airflow, dbt, AWS, BigQuery, DuckDB, Docker, LangChain, PyTorch, FastAPI, Looker, Power BI
Based in Germany, completing an M.Sc. in Information and Communication Technology at FAU Erlangen-Nürnberg.
Message me with what you are building. If it is not a fit I will say so before you spend anything.
Steps for completing your project
After purchasing the project, send requirements so Muhammad can start the project.
Delivery time starts when Muhammad receives requirements from you.
Muhammad works on your project following the steps below.
Revisions may occur after the delivery date.
Schema design and approval
Design the schema and share it for your approval, including which fields get history tracking.
Build, test and delivery
Build, test and deliver the pipeline with documentation and a walkthrough of how to run it.