MD BASHIRUN SULTANA
CS Student & ML Developer | Python, Computer Vision & LLM Agents
Hyderabad
25wh1a05r1@bvrithyderabad.edu.in
Here is a concise, well-structured paragraph that fits perfectly into the **About** section without feeling overloaded: I am a B.Tech Computer Science student at BVRIT Hyderabad seeking a Software Engineering or Machine Learning Internship. Passionate about systems programming and AI, I have built five end-to-end projects in Python—ranging from **NOVA**, a custom programming language and bytecode VM built from scratch, to an **autonomous 3-role LLM code generation pipeline** using Groq's Llama-3.3-70B. My technical background spans compiler design, computer vision (MobileNetV2), bio-data analytics (PCA & Random Forest), and signal processing, backed by an intensive 24-week in-person AI/ML program at IIIT Hyderabad. I enjoy tackling complex engineering challenges, writing clean software, and creating practical AI systems.
NOVA: Programming Language & Bytecode VM
Built a programming language from scratch with a lexer, recursive-descent parser, AST, bytecode compiler and stack-based VM. Supports functions with recursion, short-circuit logic, and line/column compiler errors, plus --dis and --trace flags. Browser demo on GitHub Pages.
: Multi-Agent Code Generation Pipeline
3-role LLM pipeline (Architect, Executor, Reviewer) on Llama-3.3-70B that generates, runs and self-corrects code with up to 4 retries. Includes an AST-based analyzer that blocks unsafe imports before execution.
15-class leaf disease classifier on the 20,638-image PlantVillage dataset using a pre-trained MobileNetV2 with augmentation. Reached 86.1% validation accuracy by epoch 3.
Attendee · Oct 24, 2026