JMC API Optimization

Problem

The mobile app backend at JMC Indonesia was slow and unpredictable. Large dataset syncs took minutes, responses wandered into the seconds, and order-sensitive workflows had no reliable ordering guarantee. The API needed to become fast, consistent, and trustworthy under real load.

Role & approach

As the backend engineer on the team I owned the API layer end to end — query tuning, data synchronization, and queue reliability.

01Mobile clientsiOS & Android apps hitting the API

→

02Laravel APIindexed queries, eager-loaded relations, cached reads

→

03MySQLstrategic indexes on hot columns

→

04Queue workerschunked batch jobs + FIFO ordering

Challenges & solutions

Response times over 200 ms. The first pass was query analysis: missing indexes, N+1 relation loads, and repeated cache-misses dominated the profile. I added strategic indexes on the hottest columns, replaced lazy relation access with eager loading, and introduced caching on read-heavy endpoints. The result was consistent sub-200 ms responses — measured, not assumed.

Large dataset synchronization. Full syncs of big tables blocked workers and sometimes failed partway, leaving inconsistent state. I switched to chunked batch processing so jobs processed data in slices, retried safely, and ran 60–80% faster while improving data reliability.

Order-sensitive workflows. Some pipelines had to process records in a strict order. I implemented FIFO queue logic so items were consumed in the exact sequence they were enqueued — no reordering, no dropped work.

Outcome

• Consistent API responses under 200 ms.
• Dataset syncs 60–80% faster with chunked batch processing.
• Production-ready applications delivered in about one week — roughly 80% faster MVPs.
• Reliable FIFO behavior for order-sensitive data workflows.