I build backend systems and offline AI tools, and design transistor-level RF circuits on the side. Most profiles pick one lane — I work both, and that combination is the throughline of what's below.

Backend & APIs

AI / ML

• RAG pipelines (retrieval-augmented generation)
• ChromaDB vector stores, sentence-transformers embeddings (all-MiniLM-L6-v2)
• LLM integration — Gemma, Ollama

Infra

• Nginx + Gunicorn/Uvicorn workers, PgBouncer connection pooling
• JWT-based auth (admin/user roles), pydantic-settings, asyncpg
• Git/GitHub workflows — branch protection, PR review, conventional commits

Electronics

• RF transmitter circuit design (AM/FM, multi-stage ~98 MHz builds)
• Transistor amplifier analysis (2N2222, BC547 — common-emitter, bias networks)
• Proteus simulation & troubleshooting

Offline AI model that tracks a student's strengths/weaknesses against the Tanzania syllabus and recommends lessons or exercises accordingly. FastAPI SQLite Gemma RAG BKT-based adaptive mastery engine

PDF-trained Q&A assistant. Ingests documents, embeds and retrieves via ChromaDB, and answers fully offline — no external API calls. FastAPI React ChromaDB Gemma / Ollama sentence-transformers

FastAPI + PostgreSQL backend built module-by-module (config, security, database, dependencies), designed for a load-balanced deployment (Nginx + Gunicorn/Uvicorn + PgBouncer). FastAPI PostgreSQL asyncpg JWT

Replace or add repo links once you decide which ones go public — see note below.

Building out the embeddings / vector-indexing module for the offline RAG chatbot — batch embedding and ChromaDB storage via scripts/build_index.py.

• GitHub: Abdully-dev
• Email: (seniorkalima194@gmail.com)
• WhatsApp: Chat on WhatsApp
• LinkedIn: Abdulsalam Kalima