PhD from Trinity College Dublin. 10+ years turning complex AI research into production systems for global clients.
Years Experience
Trinity College Dublin
Research Publications
Client Portfolio
From Academic Research to Real-World AI Impact
I bridge the gap between cutting-edge AI research and practical business solutions. With a PhD from Trinity College Dublin focused on Finding Diachronic Sense Changes by Unsupervised Methods, I bring deep theoretical knowledge combined with hands-on industry experience spanning academia and industry.
My journey took me from computational linguistics research to leading AI initiatives at EY's AI solutions delivery services team, where I architected AI solutions for international clients and managed cross-functional engineering teams. I've delivered everything from conversational analytics for early-stage startups to enterprise-scale text classification systems processing millions of documents.
Currently, I work with multiple innovative companies as an independent consultant:
I serve as a reviewer for CoLing and EACL conferences and have published extensively in top-tier AI/ML venues. My work combines academic rigor with pragmatic problem-solving, helping organizations transform unstructured data into competitive advantages through AI everywhere.
Languages: Python, R, C++, Java
ML/NLP: PyTorch, scikit-learn, NLTK, Transformers, LLMs
Agentic AI: LangChain, LangGraph, AutoGPT, CrewAI
Infrastructure: PostgreSQL, ClickHouse, FastAPI, Celery, Airflow
Two ways to engage
Your bottleneck is time, labour, or human error. I design agentic pipelines, LLM workflows, and text analytics systems that run without you.
You have an idea but no clear path to production. I architect and ship custom AI systems — from early MVP to enterprise scale.
Production systems built for real clients
A full-stack collaborative workspace where teams assemble AI personas, run multi-agent strategic simulations, and brainstorm in real time. WebSocket-driven live updates throughout.
An agentic data science assistant that interprets natural language instructions, autonomously plans multi-step transformations, and executes statistical analyses on structured data — no code, no formulas.
A FastAPI service that processes uploaded videos through Gemini's multimodal capabilities to extract key moments, insights, and generate structured analytical reports at scale.
An agentic pipeline that translates PDF documents page-by-page with cross-page context management, back-translation quality checks, and automatic retry for pages below confidence thresholds.
A multi-stage agentic pipeline that autonomously fetches captions, orchestrates translation and quality refinement via Gemini, triggers voice-cloned speech synthesis, and hands off to a video generation agent — with a human-in-the-loop review gate between stages.
A production-ready event platform with digital QR ticket generation, encrypted attendee data, camera-based live check-in, and a secure admin dashboard for event organizers.
What clients say about working with me
"When I led EY's AI Lab in Europe, I worked with Arun on core client projects for about 18 months. Thanks to his rich academic and industrial experience on natural language processing and AI, he played a pivotal role in all our AI projects, leading some of them on his own and managing project engineers. A particularly challenging project needed a full engineering solution for signing international harmonized trade codes to product descriptions in multiple jurisdictions. Arun designed the entire AI solution and executed it with his team thoroughly, achieving great success and client satisfaction. His designed architecture was then reused for tackling similar AI problems in our company. Arun is a technical leader and a great asset for any organization. I cannot recommend him enough!"
"Arun's versatility and technical depth have been invaluable across multiple challenging projects. For our non-profit initiative, he architected an automated system to translate YouTube videos from Tamil to English, making educational content accessible to wider audiences. He later helped us build sophisticated topic classifiers and a recommendation engine for our social media application. His ability to deliver production-ready AI systems that solve real-world problems is exceptional."
"Arun delivered a sentiment analysis system for our influencer marketing platform that significantly improved our campaign recommendations. His attention to detail and ability to fine-tune models for our specific domain was impressive. He also provided excellent documentation and training for our team. A true professional who delivers results."
Research contributions to the AI/ML community
Martin Emms, Arun Jayapal
COLING 2016, Osaka, Japan
Developed a Gibbs sampling algorithm for diachronic modeling to detect sense emergence from raw time-stamped n-gram data.
Martin Emms, Arun Jayapal
ICON 2015, Trivandrum, India
Martin Emms, Arun Jayapal
MWE Workshop, EACL 2014, Gothenburg, Sweden
Arun Jayapal, Martin Emms, John D. Kelleher
SemEval 2014, COLING, Dublin, Ireland
Erwan Moreau, Arun Jayapal, Gerard Lynch, Carl Vogel
PAN, CLEF 2015
Ready to transform your text data into actionable insights?