About Me
Education
University of North Carolina at Pembroke
Bachelor of Science, Major: Computer Science, Minor: Cybersecurity
Expected May 2025
Wake Technical Community College
Associates in Engineering Transfer Program, Major: Computer Science
August 2022
My Skills
I am a dedicated and passionate individual pursuing my interests in Computer Science, Data Science, and AI/ML. My drive stems from a deep love for tackling intricate challenges, whether in machine learning, data exploration and engineering, AI, web security, or development and operations. I thrive on the process of problem-solving, spending days searching for solutions, experimenting with different approaches, learning from failures, and eventually uncovering the right path. It's the journey, not just the result, that excites me most. I am a quick learner, often eager to join and explore new things.
Work Experience
My professional journey in AI research, development, and engineering.
LLM Research and Development Engineer
AI Institute of South Carolina – in collaboration with UNCP
Project: Knowledge Editing and Testing Framework for LLMs
- Researched and implemented LLM knowledge editing techniques to efficiently update models without full retraining, reducing computational costs while improving adaptability
- Designed custom augmentation pipelines and benchmarking strategies to evaluate and optimize LLM performance across diverse data domains
- Developed multi-scale AI/ML solutions with MLOps workflows using Amazon SageMaker for automated training, tuning, and deployment
- Conducted comparative analysis of LLM architectures and contributed to system change management, ensuring seamless migrations aligned with stakeholder goals
AWS SageMaker
AWS LambdaLLM Research and Development Engineer
AI Institute of South Carolina – in collaboration with UNCP
Project: Knowledge Editing and Testing Framework for LLMs
- Researched and implemented LLM knowledge editing techniques to efficiently update models without full retraining, reducing computational costs while improving adaptability
- Designed custom augmentation pipelines and benchmarking strategies to evaluate and optimize LLM performance across diverse data domains
- Developed multi-scale AI/ML solutions with MLOps workflows using Amazon SageMaker for automated training, tuning, and deployment
- Conducted comparative analysis of LLM architectures and contributed to system change management, ensuring seamless migrations aligned with stakeholder goals
All Technologies:
AWS SageMakerAWS LambdaFull Stack AI Engineer
NSA Laboratory for Analytic Sciences (LAS) @ NCSU
Project: Internet Routing Integrity & RPKI Visualization Platform
- Developed full stack application using Angular, Spring Boot, and REST APIs with real-time data visualizations for routing integrity analytics
- Implemented ETL/ELT pipelines for large-scale datasets and collaborated with data engineering to model organizational metadata for AI pipelines
- Built LLM-based classification models and an interactive RAG-powered chat interface with custom PostgreSQL pgVector database integration
- Established scalable LLM workflows including preprocessing, fine-tuning, and deployment with RayTune optimization, presenting findings at NSA-hosted conferences
Java
Flask
TypeScript
OpenAI
Claude
GeminiFull Stack AI Engineer
NSA Laboratory for Analytic Sciences (LAS) @ NCSU
Project: Internet Routing Integrity & RPKI Visualization Platform
Java- Developed full stack application using Angular, Spring Boot, and REST APIs with real-time data visualizations for routing integrity analytics
- Implemented ETL/ELT pipelines for large-scale datasets and collaborated with data engineering to model organizational metadata for AI pipelines
- Built LLM-based classification models and an interactive RAG-powered chat interface with custom PostgreSQL pgVector database integration
- Established scalable LLM workflows including preprocessing, fine-tuning, and deployment with RayTune optimization, presenting findings at NSA-hosted conferences
All Technologies:
JavaFlaskTypeScriptOpenAIClaudeGeminiDL Research and Development Engineer
Corvid Technologies @UNCP
Project: Foundational Model Development for Rapid Blood Clot Detection using DNNs
- Designed and implemented PyTorch-based DNN models for blood clot prediction with Ray Tune hyperparameter optimization
- Developed automated data preprocessing pipelines for complex CFD datasets, improving model generalization and training convergence
- Established modular ML pipelines with CI/CD integration and migrated development to AWS cloud infrastructure for improved collaboration
- Optimized inference latency and model deployment, contributing to peer-reviewed publications and delivering presentations at academic events
AWS
DynamoDBDL Research and Development Engineer
Corvid Technologies @UNCP
Project: Foundational Model Development for Rapid Blood Clot Detection using DNNs
- Designed and implemented PyTorch-based DNN models for blood clot prediction with Ray Tune hyperparameter optimization
- Developed automated data preprocessing pipelines for complex CFD datasets, improving model generalization and training convergence
- Established modular ML pipelines with CI/CD integration and migrated development to AWS cloud infrastructure for improved collaboration
- Optimized inference latency and model deployment, contributing to peer-reviewed publications and delivering presentations at academic events
All Technologies:
AWSDynamoDBMy Projects
A selection of my research and personal projects focusing on machine learning, data analytics, and web development.
DNN Models for Blood Clot Detection
Conducted research to develop foundational DNN models using PyTorch and Ray Tune for predicting blood clots (Thrombogenesis) in COVID-19 patients, analyzing Computational Fluid Dynamics (CFD) data. Explored hyperparameter optimization, dimensionality reduction, data-streaming techniques, GPU optimization and scheduling techniques, and data encoding techniques.
AWSDynamoDB
Internet Routing Integrity Project
Led a student team in a full-stack research project with the NSA Lab (LAS @ NCSU). Explored ARIN datasets, visualized RPKI information and helped develop a full-stack website for policymakers using Angular and Flask. Built ETL pipelines and integrated REST APIs. Later presented and demoed the project to NIST.
FlaskAWSTypeScriptOpenAIClaudeGeminiJava
AI-Udoin? Personal Therapy Multi Voice Agent
Developed a therapy-oriented multi-call agent (HackUNCP 2025 Healthcare Track Winner). Engineered context-aware conversations using LLMs augmented via Pinecone vector DB (RAG) and built a dynamic multi-voice synthesis pipeline. We are exploring further implimentations of this project
Car Reviews Analysis with LLMs ('Car-ing')
Built a FastAPI-based NLP service ('Car-ing') using DistilBERT, Helsinki-NLP, BART, etc. for sentiment analysis, translation, Q&A, and summarization. Deployed with Docker Compose and experimented with Azure Kubernetes. Implemented MLOps pipelines.
FastAPI
Docker
Kubernetes
1-Layer Digit Recognition from Scratch
Developed a single-layer neural network for digit recognition using only Pandas (83% accuracy) by manually implementing propagation and updates. Later version with PyTorch achieved 94% accuracy.
Get In Touch
Interested in collaborating on research, discussing projects, or exploring potential opportunities? Feel free to reach out!
Contact Information
howaisi.h@gmail.com
Phone
This feature is coming later :)
Location
North Carolina