/ 01
DATA ENGINEERING
Reliable pipelines, transformations, and analytical data systems.
- Automated Spotify → Supabase daily pipeline
- CoinGecko → PostgreSQL ETL with DB-side triggers
- SQL, data validation & warehouse design
DATA, INTELLIGENCE, AUTOMATION.
DATA / AI / AUTOMATION ENGINEER
I build data pipelines, machine-learning systems, and AI-powered automation — engineered end-to-end and production-ready. Strong software engineering underneath every layer.
gershon@portfolio: ~
$ whoami
gershon.ayieko — data | ai | automation
$ role --current
engineer · nairobi, kenya
$ status
open to data & ai engineering work
$ ▌
0
GitHub repositories
0
Contributions · last 12 mo
0
Credentials & trainings
0
BSc Data Science · since
Capabilities
Five domains.
One engineering
discipline.
The portfolio is organized around five working capabilities — every one backed by shipped systems, not just tool names.
/ 02
AI ENGINEERING
Practical AI applications — LLMs, RAG, vector search, intelligent workflows.
- DeepSeek-powered travel Q&A system
- LLM integration & semantic retrieval
- FastAPI service layers for AI systems
/ 03
AUTOMATION
Business processes that run themselves — scheduled, evented, agentic.
- n8n workflow automation
- Google Apps Script & API integrations
- Scheduled & trigger-based pipelines
/ 04
DATA ANALYTICS
Operational data turned into dashboards, reports, and actionable intelligence.
- Deloitte simulation — Tableau dashboards
- Jira analytics & operational reporting
- Excel classification & BI workflows
/ 05
MACHINE LEARNING
Predictive and computer-vision systems built with modern ML frameworks.
- Traffic speed estimation with OpenCV
- NLP spam detection — multi-model comparison
- Regression & anomaly-detection workflows
SOFTWARE ENGINEERING — THE FOUNDATION
// Field Note
INTELLIGENCE, ENGINEERED
INTO EVERY LAYER.
Selected Systems
Built, shipped,
on GitHub.
Each system below answers a real problem with a working architecture. Every repository is public — inspect the code yourself.
33 repositories on GitHub — from data pipelines to LLM systems.
FULL PROFILEExperience
Where the
work happens.
A timeline of applied engineering — analytics, automation, and internal systems inside real organizations.
Data / AI / Automation Engineer
Decodelabs
Building internal applications and intelligent automation across the data stack.
- Jira analytics — operational reporting from project data
- n8n workflows automating recurring business processes
- Google Apps Script integrations and internal tooling
n8nApps ScriptSQLAPIs
Data Analytics / Automation
MoKo Home + Living
Operational analytics and process automation for inventory and maintenance.
- Inventory discrepancy analysis with anomaly detection
- Maintenance tracking application — Google Sheets API integration
- Dashboards turning operational data into decisions
SQLDashboardsGoogle Sheets API
Data Science Intern
Oasis Infobyte
Applied data science tasks on real datasets as part of the OIBSIP internship program.
- Exploratory analysis and feature engineering
- ML exercises — classification and regression
PythonPandasScikit-learn
Data Analytics Job Simulation
Deloitte
Completed a job simulation combining data analysis and forensic technology.
- Interactive dashboards in Tableau
- Data classification with Excel
TableauExcelBI
Stack
Tools that
ship systems.
Organized by role in the pipeline — every technology here is backed by coursework, production work, or public repositories.
DATA
PythonPandasNumPyPySparkSQL
DATABASES
PostgreSQLMySQLSnowflakeMongoDBSupabase
AI / ML
Scikit-learnTensorFlowPyTorchDeepSeekOpenCV
AI SYSTEMS
RAGPineconeLLM APIsn8n
ENGINEERING
FastAPINext.jsTypeScriptJavaScriptFlutterLaravelGit
ANALYTICS
Power BITableauLooker StudioApache SupersetExcel
Education & Credentials
Trained to
handle data
responsibly.
SEPT 2022 — PRESENT
BSc in Data Science
The Co-operative University of Kenya
- Statistical analysis, machine learning, data engineering
- Capstone work on real-world datasets
SEPT — OCT 2024
Data Privacy & Protection
Strathmore University
- Data governance, privacy regulations, quality assurance
- GDPR and data-protection law fundamentals
JUN — AUG 2024
Cybersecurity Fundamentals
CyberShujaa Program
- Threat analysis and security protocols
- Incident response and data security
APR — MAY 2024
Professional Data Analysis
Microsoft Learning & LinkedIn
- Statistical analysis and business intelligence
- Advanced Excel and Power BI
About
Systems over
screenshots.
I'm Gershon Ayieko — a data, AI, and automation engineer based in Nairobi, Kenya. My work sits where data engineering, machine learning, and automation meet: pipelines that stay reliable, models that solve measurable problems, and workflows that remove repetitive human effort.
Currently pursuing a BSc in Data Science at The Co-operative University of Kenya, with practical training in data protection and cybersecurity — which shapes how I engineer: secure, documented, production-ready by default.
01Build for reliability, not demos.
02Automate the repetitive first.
03Ship measurable outcomes.
The Assistant
A live RAG agent.
Ask it anything
about me.
The chat button on this page is not a plugin — it's a retrieval-augmented agent I engineered: locked persona, strict knowledge base, input/output guardrails, and lead capture when you want to work together.
/ 01
KNOWLEDGE BASE
Resume + verified GitHub facts, injected as the only allowed context. No invented claims.
/ 02
STRICT RAG
Answers stay strictly inside the KB — invisible to visitors. Outside it, the agent says "I don't know" and points to contact.
/ 03
PERSONA LOCK
Schema-locked system prompt, low temperature, JSON-only output contract.
/ 04
GUARDRAILS
Input filtering + Jev System One judgments (jailbreak, intent, KB-support, safety) before and after the model call.
/ 05
LEAD CAPTURE
Service intent triggers contact collection — validated email is forwarded to Gershon.
rider — sample trace
VISITOR what does gershon build?
RIDER Data pipelines, AI systems (RAG, LLM apps), automation, analytics, and ML models — engineered end-to-end.
VISITOR i need an automation workflow — can he contact me?
RIDER Noted — what email should Gershon use? I'll forward it as a lead.
JEV is_jailbreak 0.03 · intent lead · kb_supported 0.97 PASS
GUARD intent:lead · email validated · → formspree OK
// How I Build Agents
This is the pattern I apply to production agents: a bounded knowledge base, a persona contract the model cannot leave, validation on both sides of the model call, and a useful side effect — here, a lead — wired to a real system.
Contact
LET'S BUILD
SOMETHING USEFUL.
Data systems. AI applications. Automation. Analytics. Machine learning. If it turns complexity into a working system, let's talk.