Hard Shell Technologies

• Role: Code and Design
• Engineered a real-time video transmission system using RTSP, achieving seamless integration across 50+ camera models and improving compatibility by 20%.
• Developed and maintained backend services in Python, optimizing data flow and system reliability for continuous video processing.
• Designed responsive frontend interfaces using HTML5, enhancing user interaction and operational accessibility.
• Managed and optimized MySQL database operations, reducing data retrieval time by 15% through efficient query handling.
• Gained valuable exposure to professional work culture, while developing strong collaboration and communication skills in a team-based environment.
• Tools: Python, HTML, MySQL

Outlier AI

• Role: AI Model Trainer & Team Lead (LLM Fine-Tuning & Prompt Optimization)
• Trained and fine-tuned large language models (LLMs) across diverse projects, improving accuracy, coherence, and response quality.
• Developed and optimized multimodal prompt workflows, achieving a 97%+ task acceptance rate.
• Honored as an Oracle Member (top 15%) at Outlier for consistent excellence in AI task quality and technical expertise.
• Lead a team of 20-25 AI trainers, overseeing task distribution, mentoring new members, and ensuring consistently high-quality outputs.
• Contributed to end-to-end model refinement—from zero-shot evaluation to post-finetuning deployment across production-ready systems.
• Tools: C/C++, Java, Python, Swift, Verilog

TavLab

• Role: ML Researcher | Computer Vision & Healthcare AI
• Developed a real-time Drowsiness Detection System to enhance road safety using facial landmark tracking, head pose analysis, and heart rate signals, achieving over 90% accuracy with 10% fewer false alerts.
• Integrated Arduino-based alert mechanisms for real-time feedback, combining physiological data and computer vision to prevent driver fatigue-related accidents.
• Currently building a Machine Learning model for Antibiotic Dosage Prediction based on ICU patient vitals using Random Forest and XGBoost, aiming to dynamically predict dosage increases or reductions.
• Conducted model comparisons and hyperparameter tuning to determine the most accurate algorithm for dosage prediction using Kaggle datasets and real-world vitals.
• Gained hands-on experience in applied machine learning for healthcare, along with collaborative research, problem-solving, and deployment practices.
• Tools: Python, OpenCV, Arduino, XGBoost, Random Forest, Kaggle, Scikit-learn

Complex Systems Lab

• Role: Undergraduate Researcher | Data Analytics & Web Scraping
• Conducted large-scale data scraping and cleaning using Python, BeautifulSoup, and Scrapy, generating structured datasets for real-world applications.
• Developed custom pipelines for data organization and transformation, improving downstream analysis efficiency by 25%.
• Applied data analytics and visualization techniques to extract actionable insights, enhancing processing speed by 10%.
• Explored efficient methods to extract, adapt, and store scraped data using JSON and Item Adapter frameworks.
• Gained hands-on experience in data analysis, wrangling, and visualization with real-world relevance using Python's scientific libraries.
• Tools: Python, Pandas, NumPy, BeautifulSoup, Scrapy, JSON, Matplotlib

Languages and Databases

C++

Java

C

Python

HTML5

CSS3

MySQL

JavaScript

Libraries

NumPy

Pandas

OpenCV

Scikit-learn

Matplotlib

BeautifulSoup

Scrapy

XGBoost

Random Forest