Hi, I'm Shrestha
CS & Math at UT Austin. I ship end-to-end software: full-stack SaaS, production AI agents, and ML research.
SM

About

I'm a computer science & mathematics student at UT Austin who ships end-to-end: full-stack SaaS with TypeScript & Next.js, production AI agents with RAG and multi-agent orchestration, and rigorous ML research. I'm currently a software engineering intern and team lead at TANTV Studios, leading 7 engineers on SyndexAI — a Next.js app over a .NET API serving 400K+ AI-summarized articles. Before that: two years of ML research at FARIL, the Foot & Ankle Research and Innovation Lab at Harvard Medical School, on deep-learning anatomical landmark detection, and summer 2026 at IBM on the Terraform Cloud Delivery Intelligence team.

Work Experience

Education

Skills

Python
JavaScript
TypeScript
Go
C++
Java
C
SQL
My Projects

Check out my latest work

I've worked on a variety of projects, from simple websites to complex web applications. Here are a few of my favorites.

Titan
GitHubSource

Titan

Production-inspired LLM inference runtime built from scratch in C++20 — tensor engine, GGUF loader, BPE tokenizer, CPU-optimized transformer inference. Zero prior C++ experience to a SIMD-optimized engine.

C++20
CMake
GoogleTest
GitHub Actions
Python
Agentify
Website

Agentify

AI agent discovery platform powered by a 5-agent LangGraph pipeline. Scrapes GitHub, Reddit, and HuggingFace; hybrid BM25 + pgvector retrieval; personalized ranked results with LLMOps observability.

Next.js
TypeScript
LangGraph
pgvector
BM25
HuggingFace
LangSmith
Prefect
Supabase
Escento
Website

Escento

Musician-creator marketplace for student collaborations. Musicians build searchable profiles; creators post gigs. In-app messaging with connection requests, blocking, and unread tracking.

Next.js 14
TypeScript
Tailwind CSS
Framer Motion
GSAP
React Three Fiber
Supabase
GitHubSource
Devpost

Olympus

Platform to control AWS cloud resources (S3, EC2, Lambda) through natural language via NVIDIA's Nemotron LLM, MCP, and Terraform.

React
Flask
Node.js
Terraform
Docker
DynamoDB
AWS
NVIDIA Nemotron
MCP
Converso
GitHubSource

Converso

Voice-driven learning platform built with Next.js and VAPI. 100+ daily interactive sessions with 50+ paying users in beta.

Next.js
TypeScript
Tailwind CSS
Supabase
PostgreSQL
Stripe
VAPI
Sentry
ContextoSolver
GitHubSource

ContextoSolver

Reproducible NLP research framework benchmarking Word2Vec, fastText, GloVe, and SVD+PPMI on word similarity and beam-search navigation. GloVe achieves a 99.8% navigation success rate.

Python
NumPy
Gensim
scikit-learn
NLTK
pandas
pytest

PopSpot

Cross-platform mobile app connecting pop-up vendors with nearby customers. Dual-role experience: vendor event management with live status and inventory; customer discovery via interactive map and feed.

React Native
Expo
TypeScript
Expo Router
React Native Maps
Reanimated
Supabase
GitHubSource

BiteSwipe

Swipe-based restaurant discovery platform using the Yelp API. Handles >10k swipes/min with optimized caching and query logic.

Next.js
React
TypeScript
Tailwind CSS
Supabase
Yelp Fusion API
Redis
Automated Landmark Detection

Automated Landmark Detection

Streamlit app for automated calcaneus landmark detection using YOLO and GU2Net deep learning. 94% successful detection rate with a clinical review interface.

Python
PyTorch
YOLOv8
Streamlit
OpenCV
Plotly
AutoHDR
GitHubSource
Kaggle

AutoHDR

Automatic lens barrel distortion correction without a lens profile, using a FeGAN flow-map architecture and a custom geometry-weighted loss.

Python
PyTorch
OpenCV
scikit-image
Real-ESRGAN
NumPy
Contact

Get in Touch

Want to chat? Send me an email or reach out on LinkedIn, and I'll respond whenever I can.

GitHub
LinkedIn