AI Engineer · Mid–Senior

Detailed Job Description

All roles AI / Applied·Bangalore, IN • BHIVE Workspace, AKR Tech Park (Kudlu Gate) • Full-time • Hybrid·5+ years (3+ years shipping LLMs in production) AI Engineer Mid–Senior Build and ship the applied AI layer — the in-product copilot, semantic search, rule-suggestion engine and structured-output features that customers use every day. LLM FeaturesRAGPrompt EngineeringEval HarnessesPython Team AI / Applied Location Bangalore, IN • BHIVE Workspace, AKR Tech Park (Kudlu Gate) Experience 5+ years (3+ years shipping LLMs in production) Tech stack • LangChain • LlamaIndex • OpenAI • Anthropic • Gemini • AWS Bedrock • Pinecone • pgvector • Qdrant • Python (async) • FastAPI • BLEU • ROUGE • BERTScore • LoRA • QLoRA • PEFT Apply for this role What you’ll do • 01 Implement, evaluate and ship LLM features end-to-end — RAG, tool-use, agents and fine-tunes • 02 Design and iterate on prompt strategies: chain-of-thought, few-shot, structured outputs, function calling • 03 Build the in-product AI copilot — answering steward questions, suggesting rules and explaining match decisions • 04 Develop evaluation harnesses with telemetry, guardrails and offline + online evals • 05 Integrate and benchmark third-party APIs (OpenAI, Anthropic, Gemini, AWS Bedrock) for cost and latency • 06 Collaborate with the ML Engineer on embedding strategies, retrieval quality and rerankers • 07 Work with backend engineers to package AI components as well-defined, observable microservices • 08 Maintain prompt and model version control with rollback capability for production AI features • 09 Document system behaviour, failure modes and known limitations for every shipped AI feature What we’re looking for • → 5+ years software engineering experience; 3+ years working directly with LLMs in production • → Strong Python — async APIs, data pipelines and clean, testable code • → Hands-on with LangChain, LlamaIndex or equivalent orchestration frameworks • → Experience with OpenAI / Anthropic / Gemini APIs including function calling and structured outputs • → Working knowledge of embedding models and vector databases (Pinecone, pgvector, Qdrant) • → Strong NLP fundamentals: tokenization, NER, relation extraction and summarisation • → Solid grasp of evaluation methodology — BLEU/ROUGE/BERTScore plus task-specific evals • → Experience shipping AI features end-to-end from prototype to production Nice to have • + Experience with B2B data platforms, MDM, entity resolution or recommendation systems • + Exposure to fine-tuning LLMs (LoRA / QLoRA, PEFT, instruction tuning) • + Familiarity with Salesforce or Databricks ecosystems