SBSapnil BasnetRésumé
An illustrated Himalayan horizon at sunrise

Sapnil Basnet

A computer science student and developer turning complex systems into useful, trustworthy tools.

Explore my work View résumé
Undergraduate Research Assistant · Texas State University GitHub Email

Selected work · 2025-2026

Systems built to make difficult decisions feel simple.

Security, retrieval, decision science, personal finance, public trust, and health. Each project starts with a messy real-world problem and ends with an interface people can act on.

Before I Deploy: Developer security tool concept illustration
01Developer security tool

Before I Deploy

A read-only Python scanner that catches risky Django deployment settings and potential hardcoded credentials without importing, executing, or uploading source code.

AST-based rules · redacted reports · deterministic exit codes
PythonASTSecurityCLI
LlamaIndex RAG: Retrieval-augmented generation concept illustration
02Retrieval-augmented generation

LlamaIndex RAG

A locally grounded assistant over LlamaIndex source, documentation, and resolved issues with transparent citations and hybrid relevance scoring.

8,440 chunks · dense retrieval · cross-encoder reranking
FastAPIQdrantOllamaStreamlit
Construction MCDM: Decision intelligence concept illustration
03Decision intelligence

Construction MCDM

An Excel-driven decision engine that derives criteria weights with AHP, ranks alternatives with TOPSIS, and cross-checks results with SAW.

Consistency checks · explainable rankings · reusable workbooks
PythonAHPTOPSISExcel
Bill Agent: Agentic finance interface concept illustration
04Agentic finance interface

Bill Agent

An AI subscription shield that scans transaction CSVs for price hikes, duplicate services, and forgotten recurring charges.

Actionable findings · savings estimates · negotiation prompts
Next.jsTypeScriptCopilotKitMotion
CharityGuard: Trust and verification concept illustration
05Trust and verification

CharityGuard

A charity verifier that combines IRS registration, financial health, scam-language signals, domain age, and disaster-exploitation checks.

Multi-signal trust score · explainable risk flags
FastAPIPythonProPublica APINLP
Shaktiman: AI fitness tracker concept illustration
06AI fitness tracker

Shaktiman

A team-built fitness assistant for body metrics, progress, workouts, and personalized diet recommendations. I designed and implemented the workout library experience.

Workout UI · exercise API schema · responsive visual library
ReactDjango RESTPostgreSQLJWT

About me

I'm majoring in Computer Science with a minor in Data Analytics at Texas State University. My background combines undergraduate research with hands-on experience in full-stack development and artificial intelligence. I enjoy applying what I learn to practical problems and building software that people find useful.

Experience

Undergraduate Research Assistant

Transformer-Based Light Field Salient Object Detection and Its Application to Autofocus Using ConvNeXt

Undergraduate Research Assistant

Sustainable and Cost-Effective Optimization of Concrete Mixtures with Equivalent Strength Using Neural Networks and Differential Evolution

IT Technician

Nepal Red Cross Society

Developed and maintained websites, managed data entry, and supported hardware functionality.

Skills

Languages

  • Python
  • Java
  • C++
  • C
  • MATLAB

AI and machine learning

  • PyTorch
  • OpenCV
  • LangChain
  • LangGraph
  • RAG Systems
  • LLMs
  • Agentic AI
  • Generative AI

Web and data

  • React
  • Django
  • FastAPI
  • PostgreSQL

Tools and platforms

  • n8n
  • Docker
  • Git
  • AWS
  • GitKraken