AI Software Engineer

Primary responsibilities

• Build the product features that have a model inside them, and own the whole path rather than the model alone
• Design the serving layer: latency budgets, batching, caching and what happens when the model is unavailable
• Build the evaluation harness before tuning anything, because without it improvement cannot be told from noise
• Handle the input as it actually arrives — CVs in every format a human can invent, multilingual and code-mixed text, documents photographed at an angle in bad light
• Instrument model behaviour in production and watch for drift rather than assuming stability
• Design the fallback for the case where the model is wrong, because it will be

Skills required

Day to day this role works with TypeScript, Python, APIs and LLMs.

• Three or more years as a software engineer, with production LLM or ML work
• You evaluate before you ship, and can show how
• Clear-eyed about where a model should not be making the decision
• Three or more years in software engineering with production machine learning exposure
• Strong Python and at least one other production language
• Model serving: latency budgets, batching, caching and behaviour when the model is unavailable
• Building an evaluation harness before tuning anything
• Handling messy real input — every CV format, multilingual and code-mixed text, photographed documents
• Designing the fallback for when the model is wrong, because it will be

Preferred

None of these are required. They are the things that would help, and a strong application is regularly missing all of them.

• LLM application experience
• Vector databases and retrieval
• Document processing pipelines

Working in AI & Machine Learning

Engineering with models in it. You build the evaluation set before you tune anything, and offline-to-online gaps are treated as the main risk rather than an inconvenience. Research is applied — the measure is whether it survives production traffic.

Deep specialisation is a career here, not a stepping stone. Research Engineer and Principal Data Scientist are terminal titles people hold for years.

About Bengaluru

Bengaluru is our largest site and where most of engineering, product, design, data and AI sit. The office is in the eastern corridor, and the majority of these roles are remote-friendly with the team meeting in person a few days a week.

The city has the deepest technology labour market in India, which cuts both ways: the work is here, and so is everyone else commuting to it. Teams are deliberate about core hours because a cross-city commute in the evening peak can take longer than the meeting it was for.