Ashvin Thadevoos

Software Engineer

Kochi, Kerala, India 1+ yrs exp 82 · Excellent

About

Software engineer with experience across full-stack web development (Python/Django, JavaScript, Angular) and applied machine learning. Led development of an explainable ML system for predicting student performance, combining ensemble modeling, SHAP-based interpretability, and a production-style dashboard. Strong foundation in data structures, model evaluation, and version control. Seeking a Software Engineering role to apply strong problem-solving skills to scalable, data-driven systems.

Skills & Expertise (27)

XGBoost Advanced

9.5/10

Years Exp

Python Advanced

9.0/10

Years Exp

scikit-learn Advanced

9.0/10

Years Exp

SMOTE Advanced

9.0/10

Years Exp

SHAP Advanced

9.0/10

Years Exp

LightGBM Advanced

9.0/10

Years Exp

Random Forest Advanced

9.0/10

Years Exp

Django Advanced

9.0/10

0.5

Years Exp

Logistic Regression Advanced

8.5/10

Years Exp

Full-Stack Web Development Advanced

8.5/10

0.5

Years Exp

CSS3 Advanced

8.0/10

0.5

Years Exp

HTML5 Advanced

8.0/10

0.5

Years Exp

Git Advanced

8.0/10

Years Exp

GitHub Advanced

8.0/10

Years Exp

Version Control Advanced

8.0/10

Years Exp

JavaScript Intermediate

7.5/10

0.5

Years Exp

Debugging Intermediate

7.0/10

Years Exp

bootstrap Intermediate

7.0/10

0.5

Years Exp

Object-Oriented Programming Intermediate

7.0/10

Years Exp

LINUX Intermediate

6.0/10

Years Exp

Java Intermediate

5.0/10

Angular Intermediate

5.0/10

Time management Critical thinking Communication Decision-making Problem Solving

Work Experience

Team of 4

EduPredict

Jan 2025 - Jan 2026

Built an end-to-end ML pipeline on the OULAD dataset predicting student pass/fail outcomes via standalone XGBoost and two soft-voting ensembles (XGB+RF+LightGBM; XGB+DT+LR+RF), achieving up to 89% accuracy and 0.96 ROC-AUC. Applied SMOTE for class balancing and a stratified 80/20 train-test split with min-max normalization to prevent out-of-distribution inputs. Integrated TreeSHAP explainability, averaging Shapley values across all base models for per-student, feature-level prediction explanations. Designed a recommendation engine mapping negative SHAP contributors to targeted interventions, and built role-based React dashboards for students and instructors. Validated the pipeline with unit, integration, and system testing (all passing); co-authored a paper under review at New Generation Computing (Springer).

Full-Stack Developer

Job Portal Web Application

Present - 2 Weeks

Collaborated with a 4-member team to build a full-stack job portal enabling employers to post openings and candidates to search and apply for jobs, using Django for backend logic and data management.

Full-Stack Developer

E-Commerce Web Application

Present - 2 Weeks

Independently designed and built a full-stack e-commerce site supporting product browsing, cart management, and checkout, from database schema through deployment-ready front end.