AYUSH THAKUR

3rd Year B.Tech (CSE) student at Netaji Subhas University of Technology

Full-stack developer and AI/ML enthusiast, passionate about building performant, user-focused web apps. Exploring the edge of GenAI, dev tooling, and visual computing.

Want to get in touch, react out via the links below!

Projects:

EcoMitra NextJS, PostgreSQL, Prisma

Built a dashboard that summarized banking statements and improved insight clarity, used by 50+ users.
Leveraged AI to enhance financial literacy with features including budget tracking, expense categorization, and investment education.

Ecomitra screengrab

Insights NextJS, PostgreSQL, Gemini/Exa

Developed an AI-powered full-stack search engine with dynamic query classification (Gemini 2.5 + Exa for web search) and intelligent context refinement.
Improved the user experience with source citations and multi-format exports (PDF, Markdown, Word, Text) to support smooth, interactive searching.

Email Client

Skin Cancer Detection Using EfficientNet PyTorch, EfficientNet, OpenCV

Developed and fine-tuned a deep learning model (EfficientNet-B4) for highly accurate skin cancer (melanoma) detection, achieving 94% accuracy, 96% recall, and a 0.98 AUC on the test set.
Leveraged transfer learning, data augmentation, and Grad-CAM visualization to build a robust and interpretable deep learning model for skin cancer detection.

Skin Cancer Detection Training

Achievements:

Innohacks'24 Winner (March 2024) - Built an optimized e-commerce logistics system with polygon-based delivery zones and a fair, buyer-verified review system.

JEE Mains (January 2023) - Achieved 99.44 Percentile with an All India Rank (AIR) of 6572 among 1.1 million candidates.

Co-Curricular Activities:

Web Dev Mentor (Google Developer Student Clubs, NSUT) - Mentored 30+ students in full-stack web dev, led hands-on sessions, and co-built a Next.js + PostgreSQL CMS that streamlined event management and cut manual work by ~50%.

Junior Council (Devcomm, NSUT) - Co-developed a gamified DSA platform to boost coding engagement and helped build Avinya'25's animated, responsive website using React and Framer Motion.