Selected product, backend, and AI work.

These are the projects that best show how I think: clear interfaces, strong system wiring, and enough technical depth that the tradeoffs are easy to see.

Across product UI, backend systems, ML research, and automation, I care most about building things that feel intentional rather than dressed up.

Hire me Download resume

selected builds

public source repos

downloadable reports

current product + AI roles

01Frontend/Public repo

Almari: E-commerce for Local Stores

Commerce UX that goes beyond a landing page.

E-commerce experience for local and handcrafted stores with catalog discovery, cart state, and account-aware checkout flows.

Built for

Built around giving smaller sellers a cleaner online buying flow than manual ordering or chat-based sales.

Why it matters

Strong signal for end-to-end product thinking across browsing, stateful cart interactions, backend wiring, and responsive polish.

Catalog browsing, filtering, and cart workflows designed for repeat shopping
React + Tailwind storefront integrated with FastAPI and PostgreSQL
Structured like a real multi-screen commerce product, not a single demo page

React · Tailwind CSS · FastAPI · PostgreSQL

Source

02Full Stack/Public repo

FromMyKitchen: Platform for Home Chefs

Multi-role product design for creators and customers.

Community-driven platform for home chefs to publish dishes, accept custom orders, and share discoverable profile pages.

Built for

Built to help home cooks turn recipes and dishes into a more direct ordering and discovery flow.

Why it matters

Useful hiring signal for content-heavy product UX: profiles, listings, search, and order-shaped interactions in one system.

Dish publishing and chef profile flows support creator-style content
Custom order journeys connect discovery with actual transaction intent
Node.js and MongoDB backend logic supports the React frontend experience

React · Tailwind CSS · MongoDB · Node.js

Source

03Machine Learning/Public repo + report

Songkhare: LSTM Recommendation Engine

ML idea connected to a usable listening workflow.

Sentiment-aware music recommendation system that uses text analysis and Spotify-backed recommendation flows to personalize results.

Built for

Explores how emotion detection can drive a recommendation experience instead of staying as a standalone model.

Why it matters

Shows I can take ML work past experimentation and attach it to an interface, API integration, and research write-up.

Text-based sentiment analysis drives mood-aware recommendation results
Spotify API integration powers song lookup and preview-friendly output
Project includes a report that documents methodology and technical choices

Python · TensorFlow · Keras · React

Source Download report

Spotify developer credentials are required for the full recommendation flow.

04Machine Learning/Public repo + report

GenerAI: WGAN Image Generation

Generative modeling, training stability, and research communication.

Research project exploring photorealistic landscape generation with a WGAN-GP training setup and a documented experimentation process.

Built for

Built to learn adversarial training in practice, from dataset preprocessing to loss behavior and output quality.

Why it matters

Hiring signal for Python and ML depth: model experimentation, technical reporting, and persistence through research-oriented workflows.

Uses Wasserstein GAN with gradient penalty for more stable training
Focuses on landscape datasets, preprocessing, and tuning tradeoffs
Repo and report together show both implementation and technical reasoning

Python · PyTorch · NumPy · Pandas

Source Download report

05Full Stack/Portfolio build

npFolio: Portfolio Management System

Data-heavy product thinking with finance-style workflows.

Portfolio management system for tracking holdings, surfacing asset insights, and keeping reporting structured across multiple positions.

Built for

Designed around the need to keep portfolio data, reporting views, and account workflows connected in one place.

Why it matters

Shows comfort with backend-heavy product work where domain modeling, dashboards, and data movement matter as much as UI.

Supports portfolio tracking, asset visibility, and structured reporting
Connects a Next.js interface to Flask, Java, and MySQL services
Demonstrates cross-stack work on a product with more logic than marketing UI

Next.js · Flask · Java · MySQL

Case-study details available on request

Public case-study and source links are not attached in this portfolio yet.

06Automation/Production work

Youanai Water: Fully Automated Social Marketing Platform

Live automation product shaped around real operational use.

Production-facing social marketing platform that automates campaign planning, creative generation, editing, and publishing workflows.

Built for

Built to reduce manual marketing work by combining content operations, automation, and creative tooling in one system.

Why it matters

Strong hiring signal for current production experience: agentic workflows, product UX, and automation logic built for day-to-day use.

Automates planning, asset generation, and publishing across the workflow
Combines Next.js, Fabricjs, Mastra, and Strapi in one product surface
Directly connected to current AI engineering work at Youanai Technologies

Next.js · Automation · Fabricjs · Mastra · Strapi

Live

Live product is public. Implementation details are not fully published in this portfolio.

Live automation product shaped around real operational use.

Next.js · Automation · Fabricjs · Mastra · Strapi

If you want the deeper walkthrough behind any of these builds, email me and I will share more detail on the architecture, tradeoffs, and implementation.

Email me Resume