Associate Software Engineer

We're hiring early-career engineers to join an AI engineering team building LLM-based systems. This role leans toward two things in particular: evaluating whether an AI/LLM system's output is correct, and modelling extracted information as structured knowledge (entities and relationships) so it can be trusted, queried, and reasoned over. If you've built something with an LLM and then asked yourself, "but how do I know if it's any good?" — this role is for you.

WHAT YOU’LL DO

• Build and extend LLM evaluation harnesses — golden datasets, precision/recall/F1, calibration, LLM-as-judge scoring.
• Design and work with knowledge graphs — entity-relationship modelling, graph traversal/queries, embeddings-based similarity and retrieval.
• Build GenAI/LLM pipelines in Python — prompting, RAG, structured output.
• Write clean, tested code; collaborate with senior engineers; document your work. Act in accordance with and as a good steward of Duck Creek Technologies mission, vision, and core values:
• Mission: To empower insurers to reimagine the future of insurance

WHAT YOU’VE DONE

Education and Work Experience:

• B.Tech/M.Tech with an AI/ML specialization, or CS/IT with strong, demonstrable AI coursework/projects with 0 to 2 years of experience
• Strong Python fundamentals; solid grasp of data structures & algorithms.
• Working understanding of LLM fundamentals: prompting, embeddings, RAG, basic evaluation concepts.
• A maintained personal GitHub profile with at least one substantive AI/LLM project (academic or personal) — not a cloned tutorial.
• Share your resume along with your GitHub profile link. Applications without a working GitHub link showing original project work will not be considered.

Good to have :

• Exposure to evaluation frameworks/metrics for generative AI (e.g. Ragas, DeepEval, or a self-built LLM-judge approach).
• Exposure to knowledge graphs / graph data structures (NetworkX, Neo4j, RDF/OWL, or a well-reasoned custom graph model in a project).
• Any agent framework (LangChain/LangGraph or similar), vector databases, any cloud platform, REST API basics (FastAPI/Flask).