About the role

## About the Role MeltPlan is building the "planning engine" for the $14 trillion construction industry — an AI system that optimizes decisions before construction begins. Founded by Kanav (co-founder of Innovaccer, a $3Bn healthtech company) and Tanmaya Kala (former Project Executive at DPR Construction). This role is a systems engineer for AI, not a prompt-wrapper: building structured (DAG-based), observable, evaluated, cost-aware, production-grade AI pipelines. ## Key Responsibilities - Architect AI workflows using DAG-based orchestration - Design structured prompt systems and agent flows - Build evaluation frameworks (automated + human-in-the-loop) - Implement observability: logging, tracing, failure analysis - Optimize token usage, latency, and cost across workflows - Design retrieval systems (embeddings, chunking, ranking) - Create guardrails and structured outputs for reliability ## Required Skills & Qualifications - Strong backend or systems engineering fundamentals - Experience building production AI systems (not demos) - Understanding of prompt design, RAG systems, evaluation pipelines, workflow orchestration, LLM monitoring/logging - Comfort debugging non-deterministic behavior ## How to Prepare for This Role ### Core Technical Topics to Master LLM API fundamentals, RAG architecture (embeddings, chunking, ranking), DAG-based workflow orchestration (e.g. Airflow/Prefect-style thinking applied to AI pipelines), evaluation harness design, token/cost optimization strategies, observability for non-deterministic systems. ### Recommended Free Learning Resources - LangChain/LlamaIndex official documentation for RAG patterns - DeepLearning.AI short courses on LLM application development (free tier) - Anthropic and OpenAI API documentation and cookbooks - Practice: build a small multi-step agent system and document its evaluation approach ### Interview Preparation - Common questions: how would you design an eval harness for an LLM pipeline, how do you make a non-deterministic system debuggable, tradeoffs between latency/cost/quality in RAG design - Coding round: expect a systems-design style discussion around an AI pipeline, possibly a take-home - Application note: this company explicitly wants your GitHub and links to shipped products, not just a resume ## How to Apply Send your GitHub, links to products you've shipped, and a short note on something you're proud of building, via the official posting link. ## Official Links & Resources - Official Job Posting / Apply: https://job-boards.greenhouse.io/meltplan/jobs/4146434009 - Company: MeltPlan