Skip to content
View rayklanderman's full-sized avatar
🎯
Focusing
🎯
Focusing

Block or report rayklanderman

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
rayklanderman/README.md

Raymond Klanderman · DevRay

Software & AI Engineer | Applied Machine Learning | Cloud & Automation

I design, build, and deploy practical software—from data pipelines and machine learning models to AI-powered products, APIs, mobile applications, and cloud services.

Portfolio LinkedIn Project Demos Contact


About

I'm a software and AI engineer working across full-stack development, applied ML, multimodal AI, agentic systems, and cloud infrastructure. My projects span healthcare, education, accessibility, civic technology, and business analytics.

I work across the engineering lifecycle: problem definition, data and architecture, implementation, evaluation, security, testing, and delivery. I value measurable results, well-defined limitations, and software people can actually use.

Two GitHub accounts, one body of work: @rayklanderman is my primary professional account; @RealDevRay is my second account, where I maintain projects including EyeSense, Luminae, and Sauti Mtaani Nairobi. Both accounts belong to me, and links below point to the original repositories.

Selected machine learning & AI work

Python pandas LightGBM XGBoost scikit-learn Decision analytics

Built a churn-prediction and profit-based targeting workflow using 100,000 telecom customer records and 113 engineered features derived from 98 source fields. The selected model achieved 0.698 holdout ROC-AUC on 20,000 records, against a 0.678 baseline. A value-aware strategy projected $49,526 monthly net profit per 100,000 customers and 74% campaign ROI under explicit assumptions—not realized returns. The project includes sensitivity analysis and a proposed randomized pilot.

Code & analysis · Technical article

Python SVD LightGBM Recommender systems

Developed a hybrid recommendation pipeline over approximately 10 million ratings, combining collaborative filtering with gradient boosting. Improved public-leaderboard RMSE from 1.21 to 0.82 through feature engineering and experimentation (a competition result, not a production metric).

Code · Case study

Probability Statistical inference Transformers Model interpretability

Completed six hands-on portfolio projects within the ongoing Probability for AI programme, spanning Bayesian reasoning, satellite image classification, generative sampling, and inference-time probability. In a GPT-2 knowledge-editing exercise, modified 1,536 of ~124.4M parameters, shifting a target completion to ~99% probability and passing 4/4 reported held-out checks.

Stanford portfolio · Rewrite GPT-2

Software products & engineering systems

Project Engineering focus
AI Health Chat · Case study Multilingual health-information PWA and WhatsApp integration; voice, image, text, tracking, and structured reports, with privacy and safety controls. Public beta; not a diagnostic service. Original AI Health Assistant won first place at the PLP Class of 2024 graduate hackathon.
TutaLearn · Case study Kenya CBC-focused learning platform with AI tutoring, assessments, progress tracking, and content workflows.
EyeSense · Case study Flutter visual-assistance prototype with offline OCR, speech interaction, accessibility features, and consent-based cloud vision. Hosted on my second GitHub account.
Luminae · Live Five-agent research-analysis platform using Mistral OCR, FastAPI, Next.js, structured outputs, and API cost controls. Second GitHub account.
Sauti Mtaani Nairobi · Live Bilingual civic-feedback platform with protected staff workflows, structured reporting, privacy safeguards, and human-reviewed AI assistance. Second GitHub account.
AutoResearcher · Demo Seven-agent research and report-generation pipeline built with Google Vertex AI ADK and Gemini.

Other projects: PostShifter Pro · Live Interviewer · Kazi Connect · SerenityAI · Codebase Genius · Weru Digital

Some products have private source code; where available, I share live applications, demonstrations, and technical case studies.

Automation & infrastructure

Self-hosted n8n — DevRay automation environment

I operate a self-hosted n8n instance for workflow automation and service integrations. This complements my application and ML work with hands-on experience in running automation tooling and connecting systems. The live instance is an operational environment, not a public demo; deployment details and sanitized example workflows can be documented separately without exposing credentials or production configuration.

I also work with REST APIs, serverless services, cloud platforms, databases, CI/CD, and application security across my projects.

Technical stack

Software & application development

Python TypeScript JavaScript Flutter React Next.js FastAPI

Machine learning & AI systems

scikit-learn LightGBM XGBoost OpenAI Gemini Mistral

Cloud, automation & delivery

Google Cloud Oracle Cloud n8n Supabase PostgreSQL Docker GitHub Actions

  • Languages & frameworks: Python, TypeScript, JavaScript, Dart, SQL; React, Next.js, Node.js/Express, FastAPI, Flutter, REST APIs.
  • ML & analytics: pandas, NumPy, SciPy, scikit-learn, LightGBM, XGBoost, SVD, Jupyter, feature engineering, model evaluation, Power BI, Tableau.
  • AI engineering: LLM APIs, multimodal inference, OCR, retrieval-augmented generation, agent orchestration, structured outputs, evaluation; OpenAI, Gemini, Mistral, Groq, and xAI integrations.
  • Cloud & operations: Google Cloud / Vertex AI, Oracle Cloud, Cloudflare Workers, n8n, Supabase, PostgreSQL, MySQL, Firebase, Redis, Docker, CI/CD, testing, API security, row-level security, and accessibility-aware design.

Professional background & continued learning

I build and deliver software independently. During a 2025 contract with the Pan-African Parliament, I also designed and developed a cross-platform Flutter/Dart Android/iOS application integrating parliamentary information and event livestreaming.

  • BSc Computer Science, University of the People — in progress.
  • Diploma in ICT, Management University of Africa — Distinction.
  • Stanford Probability for AI and GCI World 2026 — ongoing learning programmes with completed project work.
  • Additional technical development through ALX, WorldQuant University, and Power Learn Project.

Selected certifications: Google Cloud Professional Cloud Architect · Oracle Cloud Infrastructure AI Foundations Associate · Linux Foundation KCNA · BCS Foundation Certificate in Generative AI. Credentials


More: Portfolio · GitHub @rayklanderman · Second GitHub @RealDevRay · LinkedIn · Technical writing

Pinned Loading

  1. RealDevRay/luminae RealDevRay/luminae Public

    TypeScript 1

  2. CapstoneProject-Autoresearcher CapstoneProject-Autoresearcher Public

    Multi-Agent Crew for Instant, Citation-Rich Research Reports – From Prompt to LaTeX PDF in Under 10 Minutes

    Jupyter Notebook 2

  3. mrp-2026 mrp-2026 Public

    Movie Recommendation Project 2026

    Jupyter Notebook

  4. profit-targeted-churn-retention profit-targeted-churn-retention Public

    Churn prediction and expected-profit targeting for a US wireless carrier. LightGBM, ROC-AUC 0.698, $49.5k/month net profit per 100k customers.

    Jupyter Notebook

  5. devray-datalab-mcp devray-datalab-mcp Public

    Local-first MCP server for dataset analysis, preprocessing, and machine learning model training using Python and scikit-learn

    Python 1