AFRIDHO TAVADHU

You have successfully connected to the personal data terminal of Vadhh.

I specialize in building offline RAG systems, automating complex ML pipelines, and breaking things to understand how they work. This site runs on static HTML because databases are bloat and minimalism is a virtue.

[ VIEW FULL BIO ] [ DOWNLOAD RESUME ]

OPERATIONAL_CAPABILITIES

[ AI_ENGINEERING ] :: Building offline RAG systems using Llama 3, Ollama, and Vector Databases (Qdrant). Experience orchestrating LLMs with LangChain and serving results via FastAPI.
[ MLOPS_ENGINEERING ] :: Moving beyond notebooks. Specialization in Pipeline Automation using MLflow, Docker containerization, and GitHub Actions. Experience with PCA reduction and Clustering algorithms (Scikit-learn).
[ FULL_STACK_ARCH ] :: Constructing hybrid systems using Laravel 10 or Next.js 14. Mastery of Type-Safe frontends and robust SQL/NoSQL backends.

PROJECT_LOGS

[ 01 :: VAULTSEARCH_LOCAL_RAG ]

STACK :: [ Llama 3 / Qdrant / LangChain / FastAPI / Next.js ]

Mission: Privacy-First Document Search (No Cloud APIs).

Outcome: Engineered a secure, offline RAG system. Orchestrated Llama 3 (Ollama) and Qdrant for semantic search, eliminating data leaks. Built a FastAPI backend with sliding-window chunking for sub-second retrieval latency.

[ ACCESS REPO ]

[ 02 :: CUSTOMER_SEGMENTATION_ENGINE ]

STACK :: [ Python / Streamlit / PCA / Scikit-learn ]

Mission: Revenue Uplift Modeling for Accenture/Dicoding Capstone.

Outcome: Developed a clustering model identifying high-value customer segments, projecting a potential $22.4M revenue uplift. Built an interactive dashboard for stakeholders to visualize cluster traits in real-time.

[ ACCESS REPO ]

[ 03 :: TELCO_CHURN_PIPELINE ]

STACK :: [ Docker / MLflow / GitHub Actions / CI/CD ]

Mission: End-to-End ML Lifecycle Automation.

Outcome: Moved beyond "notebook data science" by containerizing the training pipeline. Integrated MLflow for experiment tracking and GitHub Actions for automated model testing and deployment, reducing manual intervention by 90%.

[ ACCESS REPO ]

[ 04 :: CONTACTLESS_ORDERING_PLATFORM ]

STACK :: [ Next.js 14 / MongoDB / NextAuth / Tailwind ]

Mission: SaaS-style Digital Menu Solution.

Outcome: Full-stack QR ordering platform with dual-interface architecture (Client vs. Admin). Features JWT-based session security, real-time cart state management, and optimized SEO rendering.

[ ACCESS REPO ]

[ 05 :: RETAIL_POS_WORKFLOW_SYSTEM ]

STACK :: [ Laravel 10 / Vue.js / MySQL / State Machines ]

Mission: Retail Transaction & Workflow Management.

Outcome: Engineered a hybrid monolith POS system handling high-volume transactions. Implemented a custom Workflow Engine to track order states (Pending -> Processing -> Quality Check) and enforce RBAC security protocols.

[ ACCESS REPO ]