Mary Macharia

Software Engineer: AI Backend Engineering | Systems Architecture | Type-Safe Frontend Architectures

As a software engineer, I build secure, scalable backend systems and end-to-end AI pipelines, from modular microservices to cross-platform interfaces that connect them to real users. I care about one thing: software people can trust, that solves real problems and drives measurable human impact.

Mary Macharia
Mary Macharia

About me

Software Engineer · Backend & AI Systems

I'm a software engineer who turned my passion for tech to became a craft. I learned the full picture of how software is made, from how a product is imagined to how it is designed, built, tested and shipped.

Along the way I found where I light up most: the backend, the systems behind the screen that must be secure, fast and dependable. I love building, and I love building things people can trust.

What I studied

  1. 01Backend DevelopmentMy focus
  2. 02Frontend Web Development
  3. 03Mobile Development
  4. 04UX Research
  5. 05UI/UX Design
  6. 06Quality Assurance
  7. 07Data and Machine Learning
  8. 08Product Management

Skills

Programming Languages

  • Python
  • SQL
  • TypeScript
  • JavaScript
  • Dart
  • HTML & CSS
  • Bash / Shell

Libraries & Frameworks

  • FastAPI
  • Flask
  • Pydantic
  • SQLAlchemy & Alembic
  • LLM APIs & SDKs (Gemini, OpenAI, Whisper)
  • RAG Pipelines & Vector Search
  • Embeddings & Semantic Search
  • Multimodal AI Workflows
  • Prompt Engineering
  • AI Agents & Tool Calling
  • LangChain / LlamaIndex
  • Hugging Face
  • Computer Vision (Image Classification, Segmentation)
  • Next.js & React
  • Flutter
  • Responsive Design
  • Pytest
  • Celery
  • WebSockets

Tools & Platforms

  • Git & GitHub
  • Docker
  • PostgreSQL & SQLite
  • Redis
  • Supabase / Firebase
  • Vector Databases (Pinecone, Chroma, pgvector)
  • Database Design (ERD, Data Modelling)
  • Postman
  • Swagger / OpenAPI
  • Playwright, Cypress & Jest
  • API Testing & Test-Driven Development
  • JWT & HMAC Authentication
  • API Rate Limiting & Security Hardening
  • REST API Design
  • Microservices Architecture
  • Data Extraction Pipelines
  • CI/CD (GitHub Actions)
  • Workflow Automation (n8n, Zapier, Make)
  • Linux
  • Vercel, Render & Railway
  • AWS / GCP Basics
  • System Design & Architecture Diagrams

Featured Projects

Things I have built, from first steps to production-minded systems.

Uzima Link

A multilingual, AI-powered health records platform that lets patients carry their medical history across hospitals in Kenya, in the language they speak. Patients describe symptoms by voice or text, and AI translates and structures them for the doctor. Access is consent-based: every facility must request permission before viewing a record.

  • AI
  • Multilingual
  • Health records
View more →

Prompt Guardian

An enterprise AI-security Chrome extension that detects and blocks sensitive data, like API keys, credentials and confidential content, before it is sent to public AI tools such as ChatGPT and Claude. It comes with admin-managed device enrollment and audit logging.

  • FastAPI
  • Next.js
  • Chrome extension
View more →

Veridex

A competitor analysis tool that puts everything you need to know about a competitor in one place. Look one up and get a consolidated view of who they are and how they position themselves, saving hours of manual research for founders, product teams and marketers.

  • Competitor analysis
  • Research
View more →
My first project

Budget Tracker

A personal finance app where users log income and expenses, set monthly budgets per category and see a dashboard of income, spending and balance, with alerts when they go over. It is the first project I confidently built end to end, and it started the building loop behind everything else here.

  • Full-stack
  • Finance
View more →

Student Progress Tracker

A full-stack study app for tracking subjects, topics, study sessions, quiz scores and goals. It shows in real time whether your pace will hit your goals, recommends what to review next based on your weakest topics, and sends a weekly email digest.

  • Full-stack
  • AI recommendations
View more →

E-Madini

An e-waste management PWA that connects producers and recyclers. I owned the AI/ML pipeline: a YOLO11 model that classifies electronic devices with 92.6% test accuracy, integrated into the FastAPI backend through a scan endpoint, alongside pickup requests, geocoding and database migrations.

  • YOLO11
  • FastAPI
  • PWA
View more →

Other Projects

More experiments in AI, retrieval and machine learning.

RAG: Detective Archive

A Retrieval-Augmented Generation pipeline that answers questions about illicit drug markets in Europe using only official EUDA reports, citing the exact file and page for every claim. It covers 10 PDFs split into 1,341 chunks, and refuses questions the archive cannot answer.

Tech stack

  • Python
  • ChromaDB
  • all-MiniLM-L6-v2
  • Groq API
  • PyMuPDF
  • Jupyter
View on GitHub ↗

Skin Lesion Segmentation

A U-Net model trained on the ISIC 2016 dataset that outlines the boundary of a skin lesion in a dermoscopic photo, served through a Streamlit app. It reached a mean Dice score of 0.884 and IoU of 0.810 on 379 held-out test images. A coursework prototype, not a medical device.

Tech stack

  • Python
  • TensorFlow / Keras
  • U-Net
  • Streamlit
  • NumPy
View on GitHub ↗

Pathway-Powered Agentic Gateway

A provider-agnostic AI orchestration gateway that sits between client apps and LLM providers such as OpenAI and Claude. Every prompt, tool call and response becomes an event in a Pathway stream, enabling replay, recovery and observability, with a watchdog that detects reasoning loops.

Tech stack

  • Python
  • FastAPI
  • Pathway
  • WebSockets
  • JWT
  • Docker
View on GitHub ↗

Other skills

Technical Writing

I love turning complicated technology into stories people can actually follow. Writing is how I make sense of what I build and what I learn, and how I pass it on.

I write about AI, new tools and where tech is heading, with one goal: someone reads it and walks away understanding something they did not before.

Read all on Medium ↗
Mary Macharia writing

Featured Writings

Jul 21

Inside Moonshot AI: Every Kimi Product, Explained

A look at the Chinese AI lab behind Kimi, and a walk through every product it has shipped.

Read more →
Jul 18

Web3 in 2026: What Survived the Hype

Web3 started as a promise to replace banks and middlemen. What actually lasted?

Read more →
Jul 9

We Used to Read Documentation. Now We Just Ask About It.

How asking AI about a tool is replacing reading its docs, and what that changes for developers.

Read more →
Jul 7

AGI and Beyond: The Fine Line Between Progress and Peril

What happens when the smartest thing on Earth is no longer us?

Read more →
Jun 18

The AI Tool 432,000 People Are Using: Open WebUI

A closer look at Open WebUI, an AI tool with hundreds of millions of downloads that few people talk about.

Read more →
Jun 16

Claude Finally Remembers You, Here's What That Actually Means

AI chatbots used to forget everything between chats. Here is what memory changes.

Read more →

Contact

Let's work together

Have a project, an idea or an opportunity? Send me a message and I'll get back to you.

Contact Mary Macharia