Selected work

Three products. Different constraints. Real production work.

These case studies focus on what I owned, the systems behind the interface, and the boundaries around confidential work.

Case study 01

Public product · 2023–Present

YouNegotiate logo
H2H Technologies logo

Technical lead

YouNegotiate

Leading architecture and delivery for consumer and creditor workflows across debt negotiation, payments, and account operations.

System route

How the product moves.

Modernized a stateful Livewire application into a stateless Laravel API and Next.js frontend, deployed as containerized services on AWS.

01

Next.js

consumer + creditor apps

02

Laravel

stateless API

03

AWS

containers + data

04

Integrations

payments + messaging

The constraint

Build a secure product that makes sensitive debt workflows understandable for consumers while giving creditor teams reliable operational control.

My ownership

I lead the architecture and delivery across consumer access, creditor onboarding, negotiation workflows, payments, messaging, administration, infrastructure, and production operations.

Delivered scope

The work that reached the product.

01

Consumer identity, access, and ecoPortal workflows.

02

Creditor onboarding, account operations, and negotiation states.

03

Payments, transactional messaging, monitoring, and security controls.

04

AWS Marketplace integration in progress with Perfsys.

Technology

Laravel APINext.jsAWS ECSRDSElastiCacheCloudFrontCloudFormationStripeTwilioBasis Theory

What I can show

Live product and architecture

Case study 02

Private education platform · 2024–2025

02

Private client work

Technical lead

Technical lead

Medical exam preparation platform

Led delivery of a live learning product across student workflows, administration, subscriptions, and AI-assisted revision.

System route

How the product moves.

A Next.js student application backed by a Laravel admin and API layer, with a dedicated FastAPI service for retrieval and generation. Dokploy and Railpack handle application delivery.

01

Next.js

student app

02

Laravel

admin + API

03

FastAPI

AI service

04

Postgres

data + Stripe access

The constraint

Bring courses, timed assessments, revision tools, billing, and AI support into one product without turning the learner or admin experience into a collection of disconnected systems.

My ownership

I led delivery across question banks, exams, video courses, flashcards, student activity, content imports, AI learning tools, subscriptions, and access control.

Delivered scope

The work that reached the product.

01

Question banks, mock exams, timers, notes, and result summaries.

02

Courses, video lessons, flashcards, targets, and revision activity.

03

AI question support, reference generation, and semantic learning tools.

04

Stripe subscriptions, entitlements, imports, and back-office workflows.

Technology

Next.js 15React 19Laravel 12Livewire 3FastAPIPostgreSQL + pgvectorStripeDokployRailpack

What I can show

Edited screens and architecture only

Case study 03

Private AI platform · 2024–2025

03

Private client work

Full-stack engineer

Full-stack engineer

Medical document intelligence platform

Built full-stack workflows that turn medical source material into searchable study context, generated questions, flashcards, and grounded answers.

System route

How the product moves.

Laravel manages the product workflows while FastAPI handles AI processing. PostgreSQL with pgvector stores source context and Redis supports queued work and caching on AWS.

01

Laravel

product workflows

02

FastAPI

AI processing

03

Postgres

pgvector retrieval

04

Redis

queues on AWS

The constraint

Make large, inconsistent source documents useful for learning while keeping generated output tied to the underlying material and giving administrators visibility into failures and cost.

My ownership

I worked across document ingestion, OCR, chunking, embeddings, vector retrieval, clustering, generation, streaming chat, queues, and operational diagnostics.

Delivered scope

The work that reached the product.

01

Document ingestion, extraction, OCR, and cleanup.

02

Semantic chunking, embeddings, pgvector, and contextual retrieval.

03

Flashcard and reference-question generation.

04

Grounded chat, streaming responses, diagnostics, and cost controls.

Technology

LaravelPythonFastAPIPostgreSQL + pgvectorRedisAWS

What I can show

Technical details only

Have a similar problem?

Bring the context. We'll work out whether I am the right fit.