HomeAI engineering lab

AI engineering lab

The technical work, ordered by what it proves. Operational agents first, then the retrieval and backend foundations they stand on.

04Voice AI agent

The conversation is not the product. The record is.

A caller speaks; the agent reasons, calls tools, checks the business knowledge it needs, and the call ends as a booking, a CRM record or a handoff to a person — not as an audio file somebody has to listen to later.

Working prototype
Voice agent architecture
Caller inbound enquiry Voice AI speech · response Intent & reasoning qualification criteria Tool calling lookup · validate · write Business action booking · record · handoff SYSTEMS IT CALLS Business knowledge Calendar CRM & lead systems Retrieval answers policy questions. Tools answer factual ones. The model decides neither.
Technology
Voice AI APIsPythonFunction / tool callingCRM integrationsRAG
Status, stated plainly

Built and run as a prototype. Not deployed as a production customer-facing line for a client.

06Foundations

Backend applications

Foundational projects rather than flagship work — the typed schemas, migrations and endpoints that everything above depends on.

Foundational build · learning projects

Library Management API

POST/booksregister a title
GET/books/{id}/availabilitylive copy status
POST/loansissue, with due date rules
PATCH/loans/{id}/returnclose and re-shelve

Expense Tracker API

POST/expensescategorised entry
GET/reports/monthlyaggregation by category
GET/budgets/{id}/varianceplanned vs actual
Stack
FastAPIPydanticSQLAlchemyPostgreSQLREST APIs