Bachelor of Science in Mathematics

CGPA: 8.27/10.0

Lead AI Researcher

• Spearheaded optimization of Retrieval-Augmented Generation (RAG) pipelines, achieving an 18% increase in LLM accuracy and a 25% reduction in latency through advanced re-ranking and semantic chunking strategies.
• Led a cross-functional team of 5 engineers to ship an AI-powered VS Code extension utilizing multi-agent workflows, driving a 30% boost in developer productivity.
• Engineered cost-efficient model fine-tuning workflows using Hugging Face Transformers and PEFT (LoRA), reducing cloud infrastructure costs by 40%.
• Collaborated with executive stakeholders to define AI strategy and presented data-driven insights that accelerated adoption of AI initiatives by 15%.

AI Researcher

• Optimized LLM inference by implementing tokenization improvements and prompt compression, increasing speed by 20% and response relevance by 15%.
• Architected scalable end-to-end ML pipelines using Python and Hugging Face, enabling production rollout within 4 weeks handling 10k+ daily queries.
• Designed and deployed enterprise-grade RAG systems for efficient retrieval across organizational datasets.

Flutter, Dart, TensorFlow Lite, ONNX, PyTorch

• Built a 100% offline Flutter-based mobile AI system to classify rice quality on low-end Android devices by optimizing a ConvNext Vision Transformer via ONNX graph surgery and FP16 quantization, reducing model memory footprint by 50% and tile input size by 75%.
• Prevented OOM crashes on MediaTek Helio CPUs using Dart background isolates and ARM big.LITTLE core tuning, enabling real-time predictions from 12MP images without UI lag on 6GB RAM devices.

Python, PyTorch, Google Earth Engine

• Built a hybrid satellite time-series classification system to predict crop types from Sentinel-2 imagery by optimizing seasonal pattern learning using PatchTST transformer embeddings fused with vegetation indices (NDVI, NDRE, SAVI).
• Achieved robust classification of irregular multivariate field data using a soft-voting ensemble of Random Forest, SVM, and Logistic Regression, achieving 0.7473 Macro F1.

LangGraph, RAG Pipelines, Python, SQL

• Built a multi-agent financial news system to reduce information overload through redundancy removal and actionable insight extraction.
• Delivered context-aware Q&A for traders and managed end-to-end development from problem definition to ML optimization.

LLMs, LoRA/PEFT, Hugging Face, Python

• Fine-tuned a medical LLM on clinical reasoning datasets to generate accurate and explainable responses using Chain-of-Thought prompting.
• Integrated real-time LLM inference into a Python backend for clinical question answering.