👋 Rohan Katyayani

AI/ML Engineer | LLMs | LoRA/PEFT | RAG | Vector DBs | Data Scientist | Building future - forward solutions with code and curiosity.

About Me

I'm an MSc Artificial Intelligence graduate with real-world experience in Data Science, GenAI, LoRA/PEFT LLMs, RAG Pipelines, Vector DBs (FAISS), iOS Development, and Machine Learning. Passionate about building human-first AI solutions. Currently seeking an AI/ML Engineer | Data Scientist role in the UK with skill worker visa sponsorship.

🚀 Featured Projects

1. FinRisk-Copilot

A production‑style system that detects suspicious transactions (tabular ML), generates bank‑tone explanations using a LoRA‑fine‑tuned LLM, and answers policy questions via RAG. Everything is tracked with MLflow, served via FastAPI, dockerized, and wired with CI + monitoring.

Designed to show senior‑level ownership across DS, LLMs (PEFT), and MLOps for a bank context.

2. Bunking-Traveller (GenAI Tourism Assistant)

LangChain LLM + RAG + VectorDB DB + HuggingFace Models assistant to help tourists explore smarter and faster.

3. Self-Learning ChatBot

A simple chatbot that learns from user interactions and stores knowledge in a JSON file. Every time you teach it something new, it remembers for future conversations. Uses Fuzzy Matching with difflab.get_close_match.

4. Pet Image Classifier

"Dog? Cat? Rabbit?" — powered by Swift + Xcode + CreateML + CoreML.

5. SeeFood

Built an image recognition app using Apple’s pre-trained InceptionV3 model to identify whether a photo is of a HotDog. A fun exercise in integrating ML into mobile inspired by Silicon Valley (The TV Series).

6. Text Sentiment Classifier

Basic NLP pipeline to classify sentiment from social media text.