Skip to content

Selected work

Anonymised product engineering outcomes.

Representative outcomes across backend performance, transactional systems and production AI workflows. Details are anonymised to protect proprietary systems.

Backend · PostgreSQL · Redis

Backend performance and reliability cleanup

Problem

A production API had become slow and harder to operate under real user traffic.

Work

Tightened data access paths, reduced unnecessary backend work and improved caching/state boundaries.

Result

Reduced p99 latency from approximately 300 ms to under 120 ms.

AI workflow · Structured outputs · Validation

Conversation-to-structured product data

Problem

Free-form customer conversations needed to become complete, usable product records without relying on operators to manually rewrite every field.

Work

Built an AI-assisted extraction workflow with structured outputs, schema validation, deterministic field rules and explicit failure handling before downstream use.

Result

Converted unstructured conversations into validated structured drafts that could be reviewed and passed into the existing product workflow.

Transactional systems · PostgreSQL · Background jobs

Auditable transactional ledger

Problem

Credit usage needed to be traceable, expiry-aware and safe against duplicate redemption.

Work

Designed locking, idempotency, expiry handling and duplicate-redemption prevention around ledger operations.

Result

Created a clearer transactional system with auditable state transitions and safer background processing.

Want an outcome like this scoped?

Share the product context, desired outcome, budget range and timeline. I will respond if it fits a focused Rails or AI workflow sprint.

Send a project brief