Leemasri Karna

CS graduate (RGUKT, 2026 | CGPA 8.8) specialising in ML model development and distributed AI systems. Interned at Viswam AI, IIIT Hyderabad, applying Domain-Adversarial Neural Networks for skin lesion classification, improving accuracy from 49% to 95.5% across 7 ISIC domain tasks via transfer learning and adversarial training. Built a distributed RAG pipeline with Tesseract OCR, FAISS, and Qwen LLM using Ray Serve across 2 nodes.

Seeking an ML Engineer role to build and ship production-grade models.

AI Research Intern at Viswam AI, IIIT Hyderabad (2025-05 – 2025-07)

• Implemented and fine-tuned a Domain-Adversarial Neural Network (DANN) using ResNet-18 on the ISIC skin lesion dataset to address domain shift between dermoscopic imaging sources via transfer learning (ham_loc_head_neck → msk_loc_head_neck).
• Improved model accuracy from 49% to 95.5% on the primary task (BA vs HA); validated across 7 domain-pair tasks with consistent gains — up to +56% improvement in some pairs.
• Analysed domain alignment using t-SNE visualisations and A-distance metrics, confirming reduced domain discrepancy after adversarial training.
• Produced evaluation reports comparing DANN against baseline classification approaches for the research team.
• github.com/planr2025/ISIC_DANN

Web Developer Intern at APSDC (2025-05 – 2025-07)

• Built Django web applications with RESTful APIs and SQL databases, supporting government digital-service workflows.
• Designed an event booking system with user authentication, form validation, and database-backed booking management.
• github.com/Leemasrik/event-booking-system

B.Tech in Computer Science & Engineering – Rajiv Gandhi University of Knowledge Technologies (RGUKT), Nuzvid (2022 – 2026)

Pre-University Course – RGUKT, Nuzvid (2020 – 2022)

Secondary School Education – Vignana Bharathi High School (2019 – 2020)