ArushBansal← Back to work

Work story

Engineering ⚙️

Jaguar Land Rover

Graduate Software Engineer

By Arush Bansal

Simulation with real-world texture

Automotive engineering lives on cycles — standardized patterns that stress efficiency, emissions, and durability. Lab results alone miss variability: acceleration habits, coasting, stop-start traffic. The module translated behavioral assumptions into repeatable simulations so teams could stress-test outcomes before hardware spent weeks in a cell.

Implementation was Python-first: clear data pipelines in, scenario parameters, and outputs engineers could compare across cycles. The work rewarded patience with domain experts — tiny changes in driver behavior swamped naive averages.

Models are only as honest as the behaviors you feed them.

How we approached the problem

Three anchors for the simulation work:

  1. 01

    Separate cycle from behavior

    Drive cycles provide structure; human parameters provide realism. Keeping them explicit made sensitivity analysis tractable.

  2. 02

    Validate against known baselines

    Every new scenario needed a reference point — otherwise you optimize for a story, not a measurement.

  3. 03

    Document assumptions

    Future you (and the next engineer) should see why a distribution looked the way it did, not just the chart.

In automotive, the model is never finished — it is only current enough to decide.

Visit JLR India on LinkedIn

Suggested next read

Everything else

48+ repos · side quests