Open to New Grad Opportunities

Hi, I'm

Ujjwal Bhatta

|

MS Computer Science Candidate (May 2026) with 2+ years of production experience building scalable SaaS platforms using Node.js, NestJS, and PostgreSQL.

📍 Vermillion, South Dakota | 🎓 University of South Dakota (4.0 GPA)

Node.js

NestJS

PostgreSQL

TypeScript

</>

fn()

Technical Skills

Technologies I use to build scalable systems

Backend Development

Node.jsNestJSExpressTypeScriptREST APIsWebSockets

Databases & ORMs

PostgreSQLMongoDBMySQLPrismaSequelize

Cloud & DevOps

AWS (EC2, S3, SES)AzureDockerCI/CD

Architecture & Design

MicroservicesSaaS SystemsSystem DesignAPI Design

AI & Machine Learning

PyTorchTensorFlowLLMsReinforcement LearningClaude APIMulti-Agent Systems

Tools & Testing

GitPostmanSwaggerMochaChai

Work Experience

Building production systems that scale

Backend Engineer

AITC International

Aug 2022 – Aug 2024

Remote (Nepal) | Flower Mound, TX, USA

▹

Led backend team of 3 developers, established TypeScript/NestJS coding standards

▹

Built SaaS visitor management system with real-time notifications and emergency broadcasts

▹

Developed live auction marketplace with concurrent WebSocket bidding rooms at sub-200ms update latency

▹

Created API documentation reducing integration time by 30%

▹

Automated deployments with Docker & CI/CD, reducing deployment time by 50%

Team Lead30% faster integration50% faster deployments

Software Engineer

Ultimodeal Online Shopping

Dec 2021 – Aug 2022

Kathmandu, Nepal

▹

Developed RESTful APIs with service layers and RBAC for e-commerce platform

▹

Utilized AWS S3 for image storage and SES for email delivery (500+ daily transactions)

▹

Integrated multiple payment systems: card payments, Khalti, eSewa

▹

Optimized database queries reducing API response times by 40%

40% faster APIs500+ daily transactions

Leadership

NSF I-Corps Entrepreneurial Lead

South Dakota, USA • Jan 2025 - Feb 2025 & Jun 2025 – Jul 2025

Conducted 40+ customer discovery interviews for AI product concepts including a legal research assistant using LLMs and a multi-instrument music transcription tool. Applied lean startup methodologies and rapid prototyping techniques.

40+

Interviews

Products

Featured Projects

Research and development work in AI, ML, and Computer Vision

Hackathon • 2nd Place, USD Ignite

PlanMyPlate — AI Meal Planning App

AI-powered meal planning app built in 48 hours — weekly plans, macro-optimized grocery lists, and cooking schedules.

PythonFastAPIReact+2

Reinforcement Learning

Policy Shaping with Uncertainty-Aware LLM for Multi-Task RL

Advanced RL framework with BERT-based LLM guidance and PPO achieving 99.2% success rate.

PythonPyTorchBERT+2

Information Storage & Retrieval

Protein Chatbot with FAISS & LLM Integration

RAG-based chatbot with FAISS vector search and Llama 3.2 for protein information retrieval.

PythonFAISSFastAPI+2

Distributed System

Evaluating Federated Learning Methods in Distributed Systems

Comprehensive analysis of federated learning for privacy-preserving ML in distributed networks.

TensorFlowPythonPyTorch+1

Machine Learning

Dental Caries Segmentation for Panoramic X-Ray Images

CNN-based model for automated caries detection in panoramic X-rays with high precision.

TensorFlowPythonDeep Learning+1

Computer Vision

Comparative Analysis of Object Detection Models

In-depth comparative study of YOLO, Faster R-CNN, SSD analyzing speed-accuracy trade-offs.

PyTorchTensorFlowPython+2

View All Projects on GitHub

Publication

IEEE CAI 2026✓ Accepted

Uncertainty-Aware LLM-Guided Policy Shaping for Sparse-Reward Reinforcement Learning

Proposed a framework combining BERT-based LLM guidance with Proximal Policy Optimization (PPO) using Monte Carlo Dropout for uncertainty estimation in sparse-reward reinforcement learning. Achieved 99.2% success rate and 2009.96 reward AUC in the MiniGrid UnlockPickup environment, outperforming Q-learning, DQN, and standard baselines.

Reinforcement LearningLLMsPPOBERTUncertainty EstimationPyTorch

View Paper