Hexon Data / Public-service systems
Real vehicles. Useful information.
Bus positions, arrival estimates and fleet analytics for local governments. Technical leadership in systems citizens and operators used.
- My role
- Co-founder & Director · Technical team lead
- Period
- 2017–July 2018
- Evidence available
- Firsthand deployment account + company/product record
01 / Context
The problem
Give PJ City Bus passengers visibility into vehicle positions and arrival estimates, and help local-government operators understand waste-truck fleet activity.
My contribution, and the team's
I led the technical team and ran data-science experiments. The bus application and fleet systems were team work. The historical record does not isolate my authorship of specific routing algorithms, telemetry services or mobile screens.
Constraints
The work had to connect a citizen-facing service with vehicle operations and institutional delivery. My account identifies government procurement cycles as a constraint on a small team; it does not provide a measured service-level or model-accuracy baseline.
02 / Decision
The consequential choice
The PJ City Bus app was delivered under the Petaling Jaya city council's identity. Vehicle tracking also supported fleet analytics. The output was a usable public service rather than only a data-science experiment; the precise model-to-production contribution remains undocumented.
03 / Delivery
What shipped
- PJ City Bus app with live positions and station arrival estimates for MBPJ
- Solid-waste vehicle tracking and analytics for Petaling Jaya and Penang
- Technical team leadership and data-science experiments
The documented outcome
During my tenure, the team shipped systems used by two local governments, according to my firsthand account. The company continued beyond my departure; its later JustNaik site identifies Hexon Data as the operator.
Limits of the record
No historical uptime, passenger-usage or ETA-accuracy dataset is published here. JustNaik is evidence of the company's later work, not a claim that I built its current product. The case does not assert that a particular machine-learning model powered the deployed arrival estimates.
04 / Evidence
Inspect the record
Continue the thread