arpan@portfolio — bash
whoami
skills --list
cat projects/
contact --open

Backend developer & data engineer building RAG pipelines, semantic search systems, and production APIs — based in Kathmandu, Nepal.

scroll
01about

I'm a backend developer and data engineer who likes turning unstructured information — documents, logs, threat feeds — into systems you can actually query.

Most of my recent work sits at the intersection of retrieval-augmented generation, vector search, and API design. I build ETL pipelines that ingest messy data, embed it, and serve it back through clean FastAPI endpoints with real session memory — not just demo scripts.

I graduated with a BSIT in AI and Machine Learning from Presidential Graduate School, and I split my time between backend engineering and applied data work — including a growing interest in cybersecurity and threat intelligence tooling.

// quick facts

LOCATIONKirtipur, Kathmandu, NP
FOCUSBackend · RAG · Data Eng
EDUCATIONBSIT, AI and MAchine Learning · 2025
LANGUAGESNepali, English
STATUSOpen to opportunities
02skills
[ backend ]
Languages & Frameworks
PythonFastAPIDjangoSQLRHEL7
[ ai/ml ]
AI & Data Systems
RAGLangChainChromaDBVector DBsOllama
[ tools ]
Tooling
Git GitHub Docker Postman Jupyter
[ human ]
Soft Skills
TeamworkProblem-solvingCommunicationAdaptability
03experience
Full-Stack Developer & Data Engineer — Waft Technology
Aug 2025 – Nov 2025
  • Developed backend APIs and integrated databases for web applications using Python and FastAPI.
  • Built ETL / data pipelines for CSV processing, document ingestion, and structured data storage.
  • Collaborated with the team to design scalable, maintainable systems and ensure smooth deployment.
Quantitative Research Job Simulation — J.P. Morgan Chase & Co. (via Forage)
Jun 2026
  • Completed a simulation focused on quantitative research methods.
  • Analyzed a book of loans to estimate a customer's probability of default.
  • Used dynamic programming to convert FICO scores into categorical data to predict defaults.
AI in Action Job Simulation — Vista Equity Partners (via Forage)
Jun 2026
  • Completed a job simulation involving prompt engineering and GenAI workflow automation for Vista's Portfolio Operations team.
  • Applied Vista's four-part prompt framework (Ask, Context, Examples, Desired Output) to synthesize customer NPS survey feedback into executive-ready insights, improving clarity and relevance of outputs by over 50% through iteration.
  • Designed and tested repeatable GenAI workflows (news aggregation, data reconciliation, qualitative insight summarization) using tools such as ChatGPT and Copilot.
  • Evaluated AI-generated outputs for accuracy, tone, and usefulness; implemented guardrails that reduced error rates and increased stakeholder usability.
04projects
Retrieval-Augmented Research Assistant01

Local RAG pipeline ingesting PDF/text documents, chunked and embedded into a vector database for semantic search. Conversational /ask endpoint with session memory, powered by Mistral:7B via Ollama.

PythonFastAPIChromaDBLangChain
Threat Intel Semantic Search & Graph Explorer02

Semantic search pipeline for cyber-threat intelligence — retrieves indicators (domains, IPs, emails, phone numbers) and documents, plus their mentions and relationships via a graph traversal API.

PythonFastAPIRAGVector DB
GutenRAG — Data-Scrape-to-RAG Pipeline03

End-to-end lakehouse system on public Project Gutenberg books, implementing a Bronze-Silver-Gold ETL flow with embeddings via Ollama (Mistral), served through a FastAPI endpoint.

PythonBeautifulSoupVectorDBRAG
Bird Detector with Cloud Storage & Gallery04

Computer-vision pipeline that captures, detects (YOLOv8), and uploads bird images to Appwrite Storage, with metadata storage and a real-time web gallery.

PythonYOLOv8AppwriteHTML
University Rank visualizer05

A lightweight, local data pipeline that extracts the official 2027 QS World University Rankings from an Excel spreadsheet, loads them into a optimized local PostgreSQL database, and streams highly responsive, memory-only interactive Altair visualizations directly to your default web browser.

PythonAltairGradio
Audio Vido Transcription06

A simple transcription service that converts audio or video files into text using OpenAI’s Whisper model.

PythonGradioHuggingFaceOpenAI
05licenses & certifications
AI Fluency: Framework & Foundations
Anthropic
Jun 2026
Claude Code 101
Anthropic
Jun 2026
Claude 101
Anthropic
Jun 2026
Digital Forensics Essentials (DFE)
EC-Council
Jun 2024
Ethical Hacking Essentials (EHE)
EC-Council
Jun 2024
King of the Hill
Presidential Cybersecurity Club
Mar 2024
Introduction to Cloud 101
AWS Educate
Nov 2023
Google Data Analytics Professional Certificate
Coursera
Jan 2022
Django
Udemy
May 2020
Python & Django
Broadway Infosys
Jul 2019
06contact

Open to backend, data engineering, and applied-AI roles. The fastest way to reach me is email — happy to talk about RAG systems, APIs, or anything in between.