Hire Aseem D. - LLM Engineer - Remote

Designed a claim evaluation benchmark using Natural Language Inference models to assess entailment between LLM-generated statements and retrieved documents, providing a quantitative metric for grounding performance. Improved grounding accuracy by 13% using Few-Shot examples, In-Context Learning, and structured reasoning templates to support verifiable, document-backed citations in LLM generations. Built and evaluated an agentic data augmentation pipeline that enriches Albert Invent’s data platform with LLM-generated structured chemical metadata, enabling support for better downstream ML and data applications.

Designed a claim evaluation benchmark using Natural Language Inference models to assess entailment between LLM-generated statements and retrieved documents, providing a quantitative metric for grounding performance. Improved grounding accuracy by 13% using Few-Shot examples, In-Context Learning, and structured reasoning templates to support verifiable, document-backed citations in LLM generations. Built and evaluated an agentic data augmentation pipeline that enriches Albert Invent’s data platform with LLM-generated structured chemical metadata, enabling support for better downstream ML and data applications.

* Integrated robust learning methods with Causal Agents, enabling quantification of causal effects b/w features. * Debugged and tested the post-processing pipeline of the demo to ensure production-ready releases.