
Hi, I'm Ochieng
Herman
Backend data engineer and automation developer. I turn production data into reliable, query-ready systems, then build the analytics, machine learning and automation on top of it, from SQL and Power BI dashboards to AI apps that cut out manual work.
About me
I'm Ochieng Herman, a backend data engineer based in Nairobi, Kenya. I spend my days moving production data into reliable, query-ready systems: pipelines, backend APIs and the automation around them. Over the last 4+ years that has meant ETL/ELT work for analytics teams, backend services for international clients, and more recently building AI applications that read documents: a multi-agent knowledge assistant over company PDFs, and DocFlow, an invoice-processing app where Claude does the reading but a person always approves the bill. I write about the systems I build on Medium, and I'm open to remote roles in backend and data engineering, analytics, and AI and automation.
Projects
DocFlow
Upload a supplier invoice PDF, digital or scanned. Claude reads it, the app checks the maths, dates, supplier and duplicates, a person approves it, and the bill is created once in Zoho Books through OAuth2.
- FastAPI
- Next.js
- PostgreSQL
- Claude API
- Zoho Books
- OAuth2
Passcode protected to limit API use. The passcode is in my application email.
Enterprise Knowledge Assistant
Hybrid RAG over company PDFs: semantic search in ChromaDB plus keyword search in Elasticsearch, merged with reciprocal rank fusion, answered by a multi-agent chain that refuses to guess.
- Python
- Kafka
- ChromaDB
- Elasticsearch
- LangGraph
- FastAPI
Retail Recommendation Platform
Streaming order events through Kafka into a Parquet data lake, cleaned with PySpark, then a recommendation model trained on the result.
- Python
- Kafka
- PySpark
- Elasticsearch
Skills
Backend
- Python
- FastAPI
- Django
- Laravel
- Express.js
- REST APIs
Data engineering
- SQL
- PL/SQL
- PostgreSQL
- SQL Server
- MongoDB
- Kafka
- PySpark
- Airflow
- dbt
- Debezium
- Elasticsearch
- Data modelling
- Data architecture
Analytics and BI
- Excel
- Power BI
- Apache Superset
- R
- Data visualization
- Reporting automation
Machine learning and AI
- scikit-learn
- XGBoost
- LightGBM
- PyTorch
- SHAP
- Claude API
- RAG
- LangChain
- LangGraph
- Multi-agent systems
- ChromaDB
Cloud and DevOps
- AWS
- Google Cloud
- Docker
- Linux
- CI/CD
- Git
Experience
Backend Data Engineer, Data Science East Africa, Nairobi
April 2023 to present
- Built ETL/ELT pipelines that move production database data into clean, analytics-ready stores.
- Wrote the SQL behind recurring reports and automated them so teams get consistent numbers on schedule.
- Built Power BI and Apache Superset dashboards that let stakeholders find answers without asking for a report.
- Documented pipeline logic, queries and dashboards so technical and non-technical teams can maintain them.
Backend Data Engineer, Upwork (remote)
September 2022 to present
- Built backend services for international clients with Laravel, Django and FastAPI, working remotely.
- Designed data models and exposed them through REST APIs for apps and reporting, with React and TypeScript frontends on top.
- Set up Linux servers and databases so projects could be deployed and reached remotely.
- Delivered a full backend solution for Owlswift Commerce, an international e-commerce business.
Articles
I write about the systems I build.
Enterprise Knowledge Assistant with Multi-Agent AI
How new documents stream in through Kafka, get indexed for both vector and keyword search, and are answered by a chain of agents that refuses to guess.
Building an Interactive Excel Dashboard for E-Commerce Product Analysis: A Case Study of Jumia
How discounts, ratings and pricing relate to customer engagement, with data cleaning, correlation analysis and ranked product tables.
Contact
I'm open to remote roles in backend and data engineering, analytics, and AI and automation. Email is the fastest way to reach me, and I'm also happy to talk on LinkedIn.
Prefer your own email app? Write to ochiengherman32@gmail.com.