👋 Hello, I'm

Arpan Pandey

I build

Software Developer specializing in AI & Machine Learning — turning LLMs, RAG, and AI agents into production-ready tools, from resume intelligence systems to stock forecasting platforms.

arpan.py
class Arpan:
  def __init__(self):
    self.role = "AI/ML Developer"
    self.skills = ["LLMs"]
    self.location = "Kathmandu, Nepal"
    self.open_to_work = True

  def get_passion(self):
    return "Building AI that matters"

Who I Am

AP
4+ AI Projects
3 ML Certs
5+ Public Repos

I'm a Software Developer specializing in AI & Machine Learning, based in Kathmandu, Nepal. I graduated with a BSc (Hons) in Information Technology from Herald College Kathmandu. My focus is on building AI systems that are actually useful — not just demos, but end-to-end applications with real data flows.

I work across the full AI stack: prompt engineering and LLM orchestration with LangChain and Groq, RAG pipelines backed by vector databases, time-series forecasting with ARIMA/LSTM hybrids, and web applications with Django and Streamlit. My projects connect these layers into deployable systems.

Currently building → deeper agentic workflows with tool-use and memory, and exploring multi-modal LLM applications.

Kathmandu, Nepal
Herald College Kathmandu — BSc (Hons) IT
🇬🇧 English 🇳🇵 Nepali 🇮🇳 Hindi

What I Work With

Programming Languages

Python SQL JavaScript HTML CSS

AI & Machine Learning

LLMs AI Agents RAG Generative AI LangChain Prompt Engineering NLP Conversational AI Vector Databases Sentence Transformers LSTM ARIMA CNN Time-Series Forecasting

Frameworks & Libraries

Django Streamlit Bootstrap Tailwind CSS Scikit-Learn TensorFlow Keras PyTorch Pandas NumPy Matplotlib

Developer Tools

Git GitHub VS Code Jupyter Notebook OpenAI API Groq API REST APIs Zapier Make (Integromat) MySQL SQLite

Things I've Built

AI Sales & Customer Support Automation Bot

Small sales teams waste hours on repetitive inquiry responses and manual lead qualification.
LLM-powered bot that handles FAQs, scores leads into Hot/Warm/Cold buckets, and auto-generates personalized follow-up emails — backed by a CRM dashboard with SQLite.
Full automation of the lead-to-email workflow; CSV export for CRM integration with existing tools.
PythonStreamlitGroq APILLMsSQLite

NEPSE AI Stock Forecasting System

Nepal Stock Exchange (NEPSE) investors lack accessible data-driven forecasting tools for local securities.
Django platform implementing ARIMA, LSTM, and a Hybrid ARIMA-LSTM model with RMSE/MAE/MAPE benchmarking and interactive dashboards for historical trend analysis.
Hybrid model outperformed standalone ARIMA and LSTM on holdout data; authenticated multi-user system with admin panel and feedback module.
PythonDjangoARIMALSTMHybrid Model

Hate Speech Detection & Text Classification

Manual content moderation doesn't scale — platforms need automated classification at volume.
NLP pipeline covering tokenization, stop-word removal, stemming, and feature extraction, then evaluated across multiple ML classifiers using accuracy, precision, recall, and F1.
End-to-end classification pipeline ready for integration as a content moderation microservice.
PythonNLPScikit-LearnTensorFlow

More on GitHub →

View All Repositories

Credentials

Mathematics for Machine Learning: PCA

Imperial College London — Coursera

Principal Component Analysis, dimensionality reduction, and eigenvector decomposition for ML preprocessing.

Verify on Coursera

Mathematics for Machine Learning: Linear Algebra

Imperial College London — Coursera

Vectors, matrices, eigenvalues, and linear transformations — the mathematical backbone of all ML models.

Verify on Coursera

Mathematics for Machine Learning: Multivariate Calculus

Imperial College London — Coursera

Gradients, partial derivatives, chain rule, and optimization — directly applicable to gradient descent and backprop.

Verify on Coursera

Relevant Coursework

Artificial Intelligence Machine Learning Data Structures & Algorithms Database Systems Software Engineering Data Analytics

Let's Connect

Have a project in mind or want to collaborate? I'm always open to interesting opportunities.