Open to opportunities

Tarun Chowdary Yegi

@tarunchowdaryyegi

Message

Applied ML and full-stack builder focused on robust, explainable security systems.

Zimbabwe

Message

What I'm looking for

I’m looking for a role building applied ML and full-stack security systems. I want to ship robust, explainable models and scalable backend services, and improve them under real-world data constraints.

I’m a Computer Science undergraduate with a strong interest in applied Machine Learning, scalable system design, and security-focused applications. I enjoy developing data-driven models and full-stack systems with an emphasis on robustness, reliability, and interpretability.

I’m comfortable working across the complete development lifecycle—from data analysis and model evaluation to backend architecture and deployment. I’m passionate about building intelligent systems that perform consistently under real-world constraints and evolving data conditions.

On the product side, I built Echo AI, an ML-Based Web Extension for URL Threat Detection using Python, FastAPI, Scikit-learn, and SHAP. I implemented real-time URL feature extraction and designed a confidence-based threat alert system, reducing phishing detection response time to under 2 seconds.

I also built CampusWhisper, a real-time college chat application with the MERN stack and WebSockets, including an event-driven architecture and a 5-minute message self-destruction mechanism to enforce zero long-term data retention. In independent research, I’m analyzing phishing URL detection robustness with adversarial perturbations, cross-dataset generalization (UCI and PhiUSIIL), and explainability via SHAP to support robust deployment.

Experience

Work history, roles, and key accomplishments

Phishing URL Robustness Study

Independent Research

Jan 2026 - Present (5 months)

Conducted robustness and explainability research for phishing URL detection models, evaluating performance degradation under adversarial perturbations and targeted feature manipulation. Built a cross-dataset generalization framework across UCI and PhiUSIIL benchmarks and used SHAP for global feature attribution to identify structural and domain-level indicators supporting phishing decisions.

Cross Dataset Generalization SHAP Explainability Model Evaluation Logistic Regression Random Forest XGBoost Distribution Shift Analysis

CampusWhisper Real-Time Chat

CampusWhisper

Developed a scalable MERN-stack real-time college chat platform with event-driven WebSockets to support high-concurrency messaging with low latency. Implemented 5-minute message self-destruction to enforce privacy-friendly zero long-term retention and deployed production-ready configuration for stable live traffic handling.

MERN Stack WebSockets MongoDB Scalability Privacy Controls

Echo AI URL Threat Detection

Echo AI

Built an ML-based browser extension that detects phishing URLs using a Random Forest pipeline and real-time risk scoring. Optimized lightweight URL feature extraction and extension workflows to reduce phishing detection response time to under 2 seconds, and added SHAP-based interpretability for threat alerts.

Phishing Machine Learning Random Forest fastAPI Feature Engineering SHAP Explainability Pandas NumPy

Education

Degrees, certifications, and relevant coursework

Indian Institute of Engineering Science and Technology (IIEST), Shibpur

B.Tech, Computer Science and Technology

Grade: 8.4 (CGPA)

Activities and societies: Projects include Echo AI (ML-based phishing URL detection browser extension) and CampusWhisper (real-time college web chat with MERN/WebSockets). Conducting a phishing URL robustness & explainability research project (manuscript under review).

Pursuing a B.Tech in Computer Science and Technology at IIEST Shibpur, with an anticipated completion in May 2028 and CGPA of 8.4.