Portfolio — B.Sc. CSIT, Tribhuvan University
Sahaj
Gyawali
A data scientist & AI/ML engineer who finds the fitted line through messy, real-world data — then ships it as software people actually use.
Based in Kathmandu, Nepal
Background & Education
About
I'm a CSIT student at Tribhuvan University specializing in data science and artificial intelligence. My work focuses on engineering end-to-end machine learning pipelines, building robust backend architectures, and deploying data-driven applications to production.
Academic path
B.Sc. CSIT — ongoing
Bhaktapur Multiple Campus
Tribhuvan University
Higher Secondary
Kathmandu Model College
Physics, Chemistry, Math, Biology
Personal Interests
Core Competencies
Fig. 3 — Field log
Experience
AI Developer
Ambition College HackFest 2025 · Kathmandu, NP
Selected as a core team member to architect and deploy a complete tech solution within a 3-day, 2-night hackathon.
- ▹Collaborated in a cross-functional team of four, leading integration between AI models and the frontend/backend.
- ▹Facilitated real-time brainstorming and rapid prototyping to pivot ideas into working features under strict deadlines.
- ▹Kept version control and coordination clean with Git, enabling smooth deploys during late-night sessions.
- ▹Shipped a production-ready prototype that demonstrated practical problem-solving under real constraints.
Competencies
Selected Projects, 2024–2026
Selected work
MediQuery AI — Medical RAG Chatbot
Developed a Retrieval-Augmented Generation (RAG) chatbot allowing medical professionals to query textbooks with precise page citations.
Implemented a semantic retrieval pipeline using Sentence Transformers for PDF chunking and embedding, integrated with Groq's Llama 3.3-70b for low-latency inference.
Engineered a modular FastAPI backend with decoupled layers for routing, RAG processing, and Pydantic validation.
Integrated SQLite for local chat history persistence and PostgreSQL for secure user authentication.
Created an automated ingestion script for seamless indexing of new PDF materials into the vector store.
Containerized the application using Docker and Docker Compose for consistent local and server deployments.
Tech stack
Telco Customer Churn
LLM-Powered LinkedIn Generator
Nepali Movie Recommendation
House Price Prediction
COVID-19 Data Analysis
Table 1 — Technical skills
Skills
Core languages
Database logic & fundamentals
Data science & viz
Pipelines & storytelling
AI / machine learning
Predictive modeling & neural nets
Engineering & web
Deployment & interfaces
Dev environment
Version control & workflow
Soft skills
Learning in public
Fig. 5 — Contact