Engineering · Full-time
AI Engineer
Ship applied AI features with evaluation built in — and the judgement to say when a model is the wrong answer.
- Experience
- 2–6 years
- Engagement
- Full-time
- Location
- Remote, India
About the role
We are careful about AI work. A feature that cannot be evaluated against a held-out set does not ship, and we tell clients when their requirement is better served by a query than a model. You would be the person holding that line, as well as the one building the thing when a model genuinely is the right tool.
What you would do
- Build applied AI features into client products — retrieval, extraction, classification, or generation depending on the problem
- Design the evaluation harness before the feature, and report honestly on what it shows
- Handle the unglamorous parts properly: data pipelines, chunking strategy, prompt versioning, and cost per request
- Set guardrails for failure modes — hallucination, prompt injection, and what the system does when the model is unavailable
- Tell a client plainly when their requirement does not need a model at all
- Keep up with a field that moves faster than the documentation
What we need from you
- 2–6 years in software engineering, with at least one AI or ML feature you took to production
- Strong Python or TypeScript, and comfort integrating with model APIs
- A real understanding of evaluation — precision and recall, held-out sets, and why a demo that looks good proves very little
- Experience with retrieval: embeddings, vector search, and why chunking strategy decides quality
- Intellectual honesty about what a model can and cannot do
Useful, not required
- Experience with agent or tool-use architectures
- Data engineering background — pipelines, warehousing, dbt
- Published writing or open-source work in the space
How hiring runs
Four steps, and a decision within a week.
Application
Send your CV and anything you have built. For engineering roles a repository or shipped product tells us more than a covering letter.
Intro conversation
Thirty minutes on what you want to work on and what we actually do. Ask us hard questions here — we would rather you did.
Technical conversation
A discussion of real problems, not a whiteboard puzzle. For engineering roles we will look at code you have written, or work through a small realistic problem together.
Decision
A clear yes or no with the reasoning, within a week. If it is a no, we will tell you why.
We have not published salary bands. Rather than print a range we would have to caveat, we will tell you the band for the role in the first conversation, before you invest time in a process. Compensation is set on the role and your experience, not on what you were paid previously.
Apply
Send us your work, not a covering letter.
Email career@laxorasoftware.com with “AI Engineer” in the subject line. A repository, a shipped product, or anything you can walk us through tells us far more than a paragraph about your strengths.