Key Responsibilities:

• Design, build, test, and iterate on prompts and prompt chains for production LLM workflows across client delivery and internal automation.
• Implement evaluation harnesses, guardrails, and regression tests so output quality is measurable and stays stable over time.
• Integrate LLM components with business systems and data — APIs, retrieval/RAG pipelines, and structured outputs.
• Partner with AI Specialists to convert solution designs into production builds, and document patterns for reuse.
• Tune for accuracy, cost, latency, and safety against real evaluation data — not vibes.
• Grow Caravel's library of reusable AI accelerators.

Preferred Qualifications:

• 2+ years building software or AI/LLM components, including hands-on experience crafting and iterating prompts for real applications.
• Working knowledge of LLM fundamentals: context windows, few-shot prompting, function/tool calling, structured outputs, and RAG basics.
• Proficiency in Python (or similar) and comfort with APIs, JSON, and version control (Git).
• A test-and-measure mindset — you evaluate output quality, you don't eyeball it.
• Clear written English and strong documentation habits.
• Ability to collaborate across time zones with an onshore/nearshore team.

Preferred Skills:

• Experience with evaluation frameworks (e.g., promptfoo, Ragas, or custom eval harnesses).
• Familiarity with embeddings, vector databases, and retrieval pipelines.
• Exposure to ERP/CRM ecosystems (NetSuite, Salesforce) or business-process automation.
• Experience with agentic frameworks and orchestration.

Type:

• Full-time

Location:

• Bengaluru

Workplace:

• Remote