Skip to content
View kpatc's full-sized avatar
13 followers · 13 following

Highlights

  • Pro

Block or report kpatc

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.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. 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
kpatc/README.md

Turning messy data into decisions that move businesses forward

👋 About Me

Data Scientist and AI Engineer who turns complex data into decisions that matter.

Three internships, 15+ production projects: diagnosed a $32.4M retail margin crisis across 504K transactions, scored the financial health of 9,618 African SMEs, forecasted electricity demand for 5 Moroccan cities, and built an NLP compliance pipeline for EU Pay Transparency regulation.

What I build: end-to-end ML systems (XGBoost, LightGBM, PyTorch, SHAP, MLflow), agentic RAG pipelines with hallucination detection and evaluation gates (Ragas, DeepEval, Langfuse), and GenAI copilots that let decision-makers interrogate live analytics in natural language.

The hardest part is never the model. It's framing the right problem and communicating the finding to someone who needs to act on it. I've trained in McKinsey's framework for exactly that: hypothesis-led problem solving, MECE issue trees, and the Pyramid Principle.

Data engineering is my foundation: Airflow, dbt, Snowflake, Kafka, Spark. I build the pipelines that feed the models, not just the models.

🔵 Currently: End-of-Studies Internship @ GeTeam Building an AI system for equal-value job identification under EU Pay Transparency Directive 2023/970

🛠 Technical Stack

Analytics & Machine Learning

scikit-learn XGBoost OpenCV

Python SQL Pandas Matplotlib Pytorch TensorFlow Scikit-learn XGBoost LightGBM Statistical Modeling
Feature Engineering A/B Testing Power BI Tableau Seaborn Business Analytics

🤖 AI & LLMs

HuggingFace LangChain CrewAI

LangChain LangGraph RAG CrewAI Vector Database LoRA
Claude Code Langfuse MCP HuggingFace SHAP Bifrost LiteLLM

⚙️ Data Engineering

Airflow Spark Databricks Airbyte Neo4j

Apache Kafka Apache Spark dbt Airflow Airbyte
Azure Databricks Delta Lake Snowflake PostgreSQL
Great Expectations Medallion Architecture

Cloud & DevOps

Prometheus Grafana

GCP Azure AWS Docker Kubernetes Terraform
Prometheus Grafana CI/CD FastAPI Flask

📊 GitHub Stats

Top Languages

GitHub Streak

Pinned Loading

  1. Telecom-Mediation-Rating-Billing Telecom-Mediation-Rating-Billing Public

    This academic project simulates a complete telecom data pipeline using Kafka, Apache Spark, and PostgreSQL, covering the full lifecycle of telecom usage data — from raw event generation to monthly …

    Python 1

  2. afripower-forecast afripower-forecast Public

    End-to-end electricity demand forecasting system for Moroccan cities multivariate time series with XGBoost, Prophet & LSTM · SHAP explainability · FastAPI · Streamlit dashboard

    Python 1

  3. PowerCo-Churn-Prediction-BCG PowerCo-Churn-Prediction-BCG Public

    PowerCo Customer Churn Prediction - Price Sensitivity Analysis A data-driven investigation into customer churn drivers with focus on price elasticity of demand as part of BCG X data science simulat…

    Jupyter Notebook

  4. STAGE2025-DabaFing STAGE2025-DabaFing Public

    A Desktop, Web, and Mobile Application for Fingerprint Classification and Ridge Counting in Fingerprints Using Reinforcement Learning

    JavaScript 1

  5. marketing-ab-intelligence marketing-ab-intelligence Public

    End-to-end A/B testing framework for marketing campaign analysis — statistical testing, power analysis, segmentation & Bayesian inference · Python · Streamlit · Executive reporting

    Python 1

  6. ds-interview-rag ds-interview-rag Public

    Agentic RAG for BCG X & McKinsey QuantumBlack DS interview prep — hybrid retrieval, conflict detection, LLM quality gating, and Langfuse observability.

    Python 1