Houston, USA · adithyareddygeeda@gmail.com · LinkedIn · GitHub

I'm an MS in Engineering Data Science candidate at the University of Houston (May 2026), with a BTech in Computer Science (AI) from SRM University. I build systems across machine learning, deep learning, NLP, and computer vision—from brain tumor detection and multilingual NLU to video action recognition and LLM-powered recommendation engines.

• Working on video action recognition (CNN+LSTM, R3D-18), multilingual NLU (XLM-RoBERTa), and LLM-based applications.
• Experienced in TensorFlow, PyTorch, prompt engineering, and end-to-end ML pipelines (Streamlit, Gradio, Flask).
• Looking to collaborate on AI/ML research, NLP, computer vision, and data-driven products.
• Ask me about deep learning, NLP, data science, Python, and building ML demos.

University of Houston — Master of Science, Engineering Data Science (May 2026)
Data Science, Probability & Statistics, Machine Learning, Big Data, Digital Image Processing

SRM University — Bachelor of Technology, Computer Science (AI) (June 2024)
Python, Deep Learning, DSA, Web Development, Soft Computing ·

• Developed LLM-powered job aggregation agent ingesting 15k+ postings/week from LinkedIn, Indeed, and Greenhouse via auto- mated scraping + semantic filtering
• Implemented company blacklist and ranking logic, decreasing irrelevant listings by ∼70%
• Rolled out full-stack system (FastAPI + React) supporting continuous ingestion, deduplication, and personalized job recommendations

• Fine-tuned XLM-RoBERTa for joint intent classification (60 classes) and slot filling (BIO tagging + CRF) on the MASSIVE dataset across 20 languages
• Achieved 86.9% intent accuracy and 71.5% slot F1, outperforming LSTM baseline by ∼7% in cross-lingual transfer
• Implemented zero/few-shot transfer to low-resource languages, demonstrating strong generalization (72–81% intent accuracy)

• Vector-based recommendation engine using LLMs across 6,810 books; 30% lower retrieval latency, 35% higher accuracy.
• Analyzed 1.2M+ words with AI-driven text classification and sentiment analysis; 92% accuracy in emotional insight detection.
• Interactive Gradio dashboard handling 10,000+ queries; 50% faster search, 40% higher engagement.

• ML-based sales prediction on 8,500+ records, 11 features, 10 retail outlets.
• Addressed 2,500+ missing values, outliers, and feature engineering; RMSE 1150.60, R² 0.53.

• SQL-based analytics for retention, churn, and revenue; found top 20% drive 70% revenue, 15% churn in Q4.
• Identified >3 unresolved tickets as 2.5× churn risk; supported targeted retention strategies.

• Deep learning pipeline for human action recognition in video (3D CNNs, CNN+LSTM).
• Flask and Streamlit apps for real-time prediction; UCF-101–style datasets and custom training.

AWS Certified AI Practitioner — Amazon Web Services (AWS)
Issued Mar 2026 · Expires Mar 2029 · Credential ID 0c22221198c04350859e16413040a1d8
Skills: Cloud Computing · AWS · AWS SageMaker · AWS CloudTrail · AWS Bedrock · Data Analysis

Microsoft Azure Fundamentals — Microsoft
Skills: Cloud Computing · Azure Services · Azure Security · IaaS · PaaS · SaaS · Azure Pricing & Support

Microsoft Power Platform Fundamentals — Microsoft
Skills: Power Apps · Power Automate · Power BI · Power Virtual Agents · Dataverse · Low-Code Solutions