I build LLM-integrated applications and ML systems that go from prototype to production — RAG pipelines, NLP models, and reinforcement learning agents. Currently completing my MSc at UCD, Dublin, after a CS degree (AI & ML specialization) in India.
- 🎓 MSc Data and Computational Science, University College Dublin (2025 – Present)
- 🧠 Background in PyTorch, RAG pipeline design, LangChain, Hugging Face Transformers (BERT, RoBERTa), and reinforcement learning (PPO, DQN, A2C)
- 🚀 Shipped two deployed GenAI applications: a clinical diagnostic platform (Gemini API, 87% accuracy) and a multimodal adaptive learning platform (Gemini + Whisper)
- 📄 Published research on algorithmic trading using reinforcement learning (PPO, DQN, A2C) with NLP-based sentiment signals
- 🏆 1st Place Overall, NodeBrew 2024 Hackathon (+ Best Accessibility & Inclusivity Award); 3rd Place, MLH TechTogether 2023; Best Use of MongoDB Atlas, MLH Waffle Hacks 2023
- 🔭 Currently exploring MLOps, vector databases, and LLM fine-tuning
- 💬 Open to AI/ML, GenAI/LLM, and Data Science roles in Ireland
- 📫 Reach me at [email protected] | LinkedIn
ML / AI
LLMs & GenAI
Data & Backend
Visualisation & Frontend
EMR Analysis & Disease Prediction — End-to-end clinical diagnostic system combining Scikit-learn for structured symptom data with Gemini API for RAG-style reasoning over unstructured EMR/lab reports. 87% classification accuracy. Deployed live on Streamlit.
FunLearn — AI-Powered Adaptive Learning Platform — Multimodal GenAI app using Gemini for adaptive quiz generation and Whisper TTS to convert content into on-demand podcasts. Built solo, concept to deployment.
Algorithmic Trading using Reinforcement Learning (Published) — RL agents (PPO, DQN, A2C) trained with PyTorch and Gymnasium, achieving 92% strategy accuracy in simulated markets, combined with a Hugging Face Transformers + NLTK sentiment pipeline over financial news. Findings published in a peer-reviewed journal.
📍 Based in Dublin, Ireland · 🔗 LinkedIn