Project Associate at Computational Data sciences at Indian Institute of Science (IISc), specializing in Machine Learningand AI. Passionate about GenAI, Data Pipelines, MLOps, and Open Source. Contributor to SciREX, an open-source Scientific Machine Learning (SciML) library.

🔭 Currently working on Foundational model for LLM and VLM ,Multi-Sensor Data Fusion (MSDF) for DRDO and IBM Granite Model research in multimodal AI.

• Worked on Clustering algorithms for signal processing.
• Built a data loader pipeline for ML models, ensuring scalability.
• Dockerized the pipeline and implemented CommitLint to enforce standardized commit messaging.
• Contributor to SciREX, enhancing regression models for scientific machine learning.
• Collaborating with IBM on fine-tuning the Granite Model for scientific response generation.

• Developed a Retrieval-Augmented Generation (RAG) model with multimodal architectures for improved response generation.
• Designed a scalable system using Celery & Redis, reducing processing time by 40%.
• Built a data preprocessing pipeline and web scraper using Scrapy, handling 500K+ records and 50GB+ raw data.

• Bachelor of Engineering in Computer Science (2020 - 2024)
  • Sathyabama University, Chennai
  • GPA: 8.0/10

Skills: PyTorch | ONNX | TensorRT | PyCUDA | CUDA | Model Conversion | GPU

• Developed an optimized object detection pipeline by converting YOLOv5 models from PyTorch to ONNX and further accelerating with TensorRT FP16 precision.
• Achieved 1.9× speedup over PyTorch (495.3 FPS vs 265.9 FPS) and 5.2× speedup over ONNX Runtime (495.3 FPS vs 94.9 FPS) on NVIDIA RTX A6000 GPU.
• Reduced inference latency by 46% (3.76ms to 2.02ms) through efficient CUDA memory management and asynchronous execution.
• Implemented GPU-accelerated processing with proper stream synchronization and benchmark methodology, resulting in 81% reduction in processing time compared to ONNX Runtime.
  🔗 Repo: GitHub

Skills: GenAI | Phi-2 SLM Model | Prompting | FastAPI | React | JavaScript | TailwindCSS

• Built an AI chatbot using Small Language Models (SLMs) for efficient code generation.
• Designed a React-based chat interface and integrated XML output functionality.
• Developed a data-agnostic architecture supporting easy model swaps (tested with Mistral 7B).
  🔗 Repo: GitHub

Skills: Computer Vision | OpenCV | FER | FastAPI | Python

• Built an AI-powered video analysis system to detect facial emotions in meeting recordings.
• Implemented face detection using OpenCV & Haar Cascade Classifier.
• Analyzed sentiments (anger, happiness, sadness, etc.) and visualized engagement trends.
  🔗 Repo: GitHub

Skills: IBM DocLingo | Data Extraction | OOP Principles | Logging | GPT-2

• Built a data preparation pipeline for fine-tuning LLMs on scientific documents.
• Extracted images, formulas, and captions using IBM DocLingo and proximity-based captioning.
• Designed a modular OOP-based architecture for scalability and debugging.
  🔗 Repo: GitHub

Skills: FastAPI | Celery | Redis | Docker | Concurrency

• Developed a FastAPI-Celery-Redis Orchestrator for scalable task execution.
• Implemented asynchronous task queuing to optimize AI & RAG model performance.
  🔗 Repo: GitHub

✅ Foundation of Cloud Edge Machine Learning – NPTEL (April 2023)
✅ AWS Professional Certificate – Starts Cloud (Dec 2023)
✅ DevOps Course – IIT Bombay (Sept 2022)

Skills: PyPI Packaging | Regression Modeling | Documentation | Open Source Contribution

• Developed the entire regression module for the ML section of SciREX.
• Implemented PyPI installation support and wrote comprehensive documentation.
  🔗 Repo: GitHub

📍 IISc, Bangalore | Dec 2024

• Organized India's First International Scientific Machine Learning Conference with 300+ participants.
  🔗 More Info

📍 Sathyabama, Chennai | Oct 2023

• Curated technical quiz questions and managed participant engagement for 100+ attendees.
  🔗 More Info