ace User

I’m a Python developer and an AI/ML Master’s student at Osmania University, focused on turning messy data and research ideas into working systems. My recent internship experience strengthened my ability to clean, validate, and structure real-world datasets.

At InI Labs, I collaborated with the Data Curation Team to clean and validate raw data, and I developed Python scripts that convert unstructured text into organized HTML-tagged output for better downstream usability. I also optimized existing code for faster processing, reducing execution time and improving efficiency.

In my projects, I’ve worked across the LLM lifecycle—from fine-tuning to evaluation. I fine-tuned Qwen 2.5 Coder 32B with QLoRA to improve code quality and achieved a 3.35% absolute pass@1 improvement on BigCodeBench-Hard, while designing resource-efficient training on a single H100 using Unsloth, BitsAndBytes, and Modal.

I also built an AI-powered legal research RAG system using LangChain-style components, ChromaDB embeddings, and context-aware answer generation, comparing embedding models and evaluating faithfulness and relevance. On the product side, I built and deployed an AI Stories generator using a FastAPI backend, Gemini image generation, Firebase services, and Flutter, reaching 4,000+ organic downloads.