About

I am an applied scientist at Thomson Reuters, where I develop evaluations for large language model systems that revise legal contracts and answer legal research questions.

Before that, I worked at Amii on wâsikân kisêwâtisiwin, a writing assistant developed with pipikwan pêhtâkwan. During my M.Sc. at the University of Alberta, I studied how large language models perform in automatic essay scoring.

Work

Relevant experience

Jan 2026 to Present

Remote

Applied Scientist

Thomson Reuters · Toronto, Canada

  • Built evaluation pipelines to determine whether AI-generated legal contracts correctly implement structured revision requests.
  • Developed an evaluation framework for a Legal Deep Research Agent, covering accuracy, completeness, and factual grounding.
  • Benchmarked LLM-based judges across evaluation quality, cost, and runtime.

Sep 2024 to Sep 2025

Edmonton, Canada

Machine Learning Researcher

Alberta Machine Intelligence Institute (Amii)

  • Built the backend for wâsikân kisêwâtisiwin, which detects, explains, and rewrites hateful or racist language toward Indigenous Peoples.
  • Created synthetic training data and an internal benchmark, and worked on supervised fine-tuning of Llama and DeepSeek models.
  • Designed a RAG pipeline using dense retrieval, multi-query expansion, and re-ranking.
  • Deployed the backend on Google Cloud using FastAPI and vLLM.

Jan 2022 to Jul 2024

Edmonton, Canada

Graduate Researcher

University of Alberta

  • Researched and benchmarked large language models for automatic essay scoring using few-shot prompting, instruction fine-tuning, and statistical analysis.
  • Taught Python labs and mentored a team of 8 to 10 teaching assistants.

Research

Publications

  1. Are Large Language Models Good Essay Graders?

    Anindita Kundu and Denilson Barbosa

    arXiv preprint · 2024

    Paper

  2. Alignment-free Phylogenetic Tree Estimation

    Anindita Kundu, R. T. Usha, N. K. Samia, and M. M. Rahman

    i-PACT · 2019

    Paper

  3. Automated Railway Maintenance System in Context of Bangladesh

    N. K. Samia, R. T. Usha, Anindita Kundu, et al.

    ICASERT · 2019

    Paper

Background

Education

2015 to 2019

B.Sc. in Computer Science and Engineering

Military Institute of Science and Technology · CGPA 3.89/4

Proposed an alignment-free algorithm in bioinformatics for constructing phylogenetic trees.