Janak Kapuriya

AI/ML Engineer &
Applied Researcher

I am an AI/ML Software Engineer and Applied Researcher specializing in Natural Language Processing (NLP), Generative AI, and Large Language Models (LLMs). With an M.Tech in CSE with a specialization in Artificial Intelligence and hands-on experience in end-to-end ML development, I bridge the gap between cutting-edge AI research and production-ready industrial applications.

Location: Galway, Ireland (Open to Relocate)
Legal Status: Stamp 4 Holder
Work Preference: On-site, Hybrid, or Remote

Research Assistant (AI Engineer)
University of Galway | Sep 2024 - Present

Architected and developed a Multi-Agent System using PydanticAI and Python to automate and orchestrate complex medical claim approval workflows. .
Designed a robust Retrieval-Augmented Generation (RAG) pipeline integrating LLaMA3 and PostgreSQL to retrieve relevant clinical data, ensuring agent decisions were highly accurate and contextually grounded.
Collaborated with a cross-functional team to deploy the end-to-end multi-agent solution onto Microsoft Azure, delivering a scalable and reliable production environment for automated task execution.
Engineered a robust data ingestion pipeline to process complex, unstructured business documents (PPTs, PDFs, and Images) by leveraging Vision-Language models (Pixtral-11B).
Leveraged state-of-the-art open-source LLMs (Mistral-7B, LLaMA3) to analyze the processed documents, successfully extracting key entities, topics, and complex semantic relationships to drive key business decisions.
Refined the extraction logic and utilized few-shot prompting to elevate system accuracy to 90%, transforming raw, unstructured data into actionable insights for enterprise decision-making.
Published research papers at top AI conferences and workshops (SIGIR, ICLR, ECIR).

Research Intern
National Institute of Informatics (NII Japan) | Mar 2024 - Apr 2024

Engineered a novel Semantic Frame Aggregation-based Transformer to automate live comment generation from streaming videos.
Outperformed state-of-the-art baselines, achieving a 4% performance improvement in Recall@1 on the LiveChat dataset.
Work published in the Q1 journal IEEE Transactions on Multimedia (Impact Factor: 9.7)

Masters in Computer Science and Engineering with Specialization in AI
Indraprastha Institute Of Information Technology Delhi (IIIT Delhi) | Aug 2022 - June 2024
GPA: 8.27 / 10.00
Bachelor of Engineering in Computer Engineering
Gujarat Technological University (GTU) | Aug 2017 - June 2021
GPA: 9.32 / 10.00

Multi-Agent Systems & Agentic Workflows: Pydantic AI, LangChain, LangGraph
Enterprise RAG Pipelines & Vector Databases: Pinecone, ChromaDB, FAISS
Open-Source LLM Fine-Tuning & Alignment: Deepseek, Qwen, Mistral, LLaMA3, RLHF/PPO/DPO
LLMs APIs: OpenAI, Claude, Groq, Deepseek, Openrouter, LLaMA
End-to-End ML Deployment: Docker, Microsoft Azure, PySpark, CI/CD pipelines, Flask REST APIs

TopoBench: Benchmarking LLMs on Hard Topological Reasoning
Mayug Maniparambil, Nils Hoehing, Janak Kapuriya, Arjun Karuvally, Ellen Rushe, Anthony Ventresque, Noel O'Connor, Fergal Reid
Logical Reasoning of LLMs @ ICLR 2026
Preprint

Semantic Frame Aggregation-based Transformer for Live Video Comment Generation
Anam Fatima, Yi Yu, Janak Kapuriya, Julien Lalanne, Jainendra Shukla
IEEE Transaction on Multimedia 2025
paper

Exploring the Role of Diversity in Example Selection for In-Context Learning
Janak Kapuriya, Manit Kaushik, Debasis Ganguly, Sumit Bhatia
SIGIR 2025 | Special Interest Group on Information Retrieval
paper

MM-PhyQA: Multimodal Physics Question-Answering with Multi-image CoT Prompting
Avinash Anand Janak Kapuriya, Apoorv Singh, Jay Saraf, Naman Lal, Astha Verma, Rushali Gupta & Rajiv Shah
PAKDD 2024 | Pacific-Asia Conference on Knowledge Discovery and Data Mining
paper

Winter 2024: Teaching Assistant for CSE508: Information Retrieval (IIIT-Delhi)
Monsoon 2022: Teaching Assistant for CSE201: Advance Programming (IIIT-Delhi)

Template: Ashish Sharma