Built custom Vision and Text Transformers from scratch for cross‑modal retrieval between product images and descriptions. Implemented contrastive learning for image‑text matching.
I design production‑grade
AI systems with precision.
MS in Artificial Intelligence @ DePaul (GPA 3.8). I build end‑to‑end ML solutions — from data pipelines and model training to deployment, monitoring, and iteration.
// Profile & stack
What I bring to an AI team.
I work at the intersection of machine learning, software engineering, and cloud infrastructure — turning research ideas into reliable, production‑ready systems.
Recent projects include OCR‑driven RAG workflows at Aries View Inc., human activity recognition research at DePaul University, and custom vision‑language models for cross‑modal retrieval.
Languages
ML / DL
Cloud & MLOps
AI focus
Data & vectors
Tools
// Experience
Where I've applied this in practice.
- Designed OCR + RAG workflows for document analysis, reducing manual review time by ~60%.
- Built data pipelines on AWS (S3, EC2) to support training and deployment at scale.
- Fine‑tuned and evaluated LLMs for enterprise use‑cases, improving predictive accuracy by 15%.
- Shipped ML services with monitoring, logging, and rollback‑friendly deployment patterns.
- Developed computer vision models for human activity recognition from wearable sensor data.
- Processed time‑series data from 120+ patients and engineered features for robust generalization.
- Improved model accuracy by ~14% through custom interpolation and signal‑processing techniques.
- Collaborated across CS and health research teams to prepare publications.
- Built Python automation tooling that cut repetitive workflows by ~40%.
- Collaborated with engineering teams to deliver production‑quality features on schedule.
- Helped implement CI/CD pipelines to streamline deployments.
// Selected projects
A sample of recent work.
Developed autonomous NLP system with vector‑based knowledge retrieval using RAG architecture. Implemented LangGraph for agentic workflows and ChromaDB for semantic search.
Research project using time‑series analysis and deep learning to classify activities from wearable sensors. Developed novel interpolation techniques improving accuracy by 14%.
End‑to‑end sentiment analysis with BERT‑based models, extensive preprocessing, and comprehensive evaluation. Extended with NER and topic modeling capabilities.
// Background
Academic context and credentials.
MS, Artificial Intelligence
BE, Computer Engineering
Certifications
// Contact
Let's connect.
I'm currently open to full‑time opportunities in AI/ML Engineering, Research, and Data Science roles. For collaborations or research‑driven projects, email is best.