Updated 19h ago

Conversational Agentic & RAG Engineer - Mumbai at ClanX · AI Job Signal

As an Conversational Agentic & RAG Engineer at Eighty Days, you’ll build and own LLM-powered features for travel discovery, from conversational experiences and recommendations to reliable AI pipelines on a TypeScript backend.

Company Details
Eighty Days is building a map-based travel discovery platform centred on taste, trust, and personal context. It brings recommendations from social media, search, and private conversations into one place where people can save, organise, and revisit places that matter.

Website: https://eightydays.ai/

Requirements

3-5 years of full-time software engineering experience, with strong TypeScript/Node.js skills (NestJS preferred).
Shipped and owned a conversational, search, or RAG feature used by real users at scale: a customer-facing product, not an internal tool, company knowledge-base bot, or internal RAG.
Hands-on production experience with LangChain or LangGraph for building agentic or multi-step LLM workflows.
Hands-on experience with Langfuse for tracing, evaluations, and regression testing of prompt and model changes.
Search relevance experience: built or tuned search with Elasticsearch/OpenSearch and vector search (Qdrant, Pinecone, etc.), and measured relevance rather than eyeballing it.
Data-driven approach: comfortable defining success metrics, working with product analytics, and using funnel and conversion data to decide what to build or improve.

Nice to have

Built an LLM-as-judge setup or relevance test set that changed a real decision.
Experience with geo/places data, entity resolution, or local search.
Experience with reranking and rank fusion.
Experience with messaging platform APIs (WhatsApp Cloud API, Instagram Graph API).
Experience with prompt caching or model routing at scale.
Exposure to multimodal understanding (video, images).
Exposure to Model Context Protocol (MCP).
Prior experience in a fast-paced, AI-driven startup environment.

Responsibilities

Build agentic, multi-step LLM workflows: orchestration, tool calling, memory, and multi-turn conversation state.
Own our conversational AI end to end across in-app chat: from product requirement, to production, to iteration based on real user behavior.
Own retrieval quality for our hybrid search and RAG systems, grounding every answer in first-party data.
Work on query understanding and entity resolution for location-aware search.
Make LLM features reliable in production: structured outputs, streaming, fallbacks, and routing across multiple models based on quality, latency, and cost.
Build async, event-driven pipelines that hold up under real user traffic.
Build the evaluation and observability layer: tracing, eval datasets curated from real user conversations, regression testing for prompt, model, and retrieval changes, and metrics for relevance and groundedness.
Optimise cost and latency across LLM workloads.
Identify the data a feature needs, design and architect the system around it, and source it from the right teams: product analytics, conversion and growth metrics.
Translate requirements and product insights into shipped features, collaborating closely with the founder, engineering, and product.

Job Details
Mumbai, work from office.

Interview Process