

AI Engineer with a Master's in Artificial Intelligence from Leiden University and 4+ years of experience at ISRO. Skilled in computer vision, NLP, and deep learning, with hands-on expertise in TensorFlow, PyTorch, Hugging Face, and cloud platforms. Developed mission-critical systems in aerospace and healthcare, including CT-based aorta segmentation for surgical planning. Passionate about ethical, real-world AI solutions and collaborating with global, remote teams.
Upwork
April 2024 - Present
• Worked extensively with Large Language Models (LLMs) to train and optimize AI-generated outputs. • Developed real-time object detection and tracking solutions for industrial and commercial applications. • Evaluated and trained responses for LLMs using Reinforcement Learning with Human Feedback (RLHF) as AI trainer. • Built custom AI applications for clients using Python, TensorFlow, PyTorch, and OpenAI APIs.
Surgical Reality
July 2023 - March 2024
• Designed a deep learning pipeline for thoracic aorta segmentation in DICOM images using U-Net and Transformer architectures, achieving a Dice score of 94.6%. • Collaborated with Leiden University and Erasmus MC cardiac surgeons to integrate AI models into clinical workflows. • Built AR-based 3D visualization modules to enhance pre-surgical planning for TEVAR. • Applied advanced preprocessing and hyperparameter tuning to optimize model performance.
Indian Space Research Organisation (ISRO)
August 2017 - August 2022
• Developed AI-driven telemetry format validation software using CI/CD pipeline and MLOps to verify sensor file formats, reducing errors by 30%. This required critical thinking to troubleshoot complex issues with the embedded system's data pipeline and ensured accuracy for 17 launch campaigns. • Designed Python-based automation tools for data analysis of upper stage electronics, leveraging machine learning algorithms to reduce review time by 45% and enhance troubleshooting. • Published telemetry format validation research in the “Journal for Aerospace Quality and Reliability”. • Enhanced pre-launch and post-flight data analysis for the Chandrayaan mission using machine learning algorithms, ensuring data precision and mission reliability.
NRSC (ISRO)
June 2016 - July 2016
• Optimized machine learning models for land feature classification on ASTER geological datasets, achieving a 98% accuracy through advanced hyperparameter tuning. • Conducted a comparative study of Backpropagation Neural Networks and SVM, improving feature extraction accuracy in geospatial imaging.
Defence Research & Development Organisation (DRDO)
November 2015 - December 2015
• Designed and developed Robust Roll Autopilot using Extended State Observer in MATLAB to control roll movement in tactical missile.
Leiden University, March 2024
Computer Science
Indian Institute of Space Science and Technology, March 2017
Avionics
DeepLearning.AI
Issued: 5/1/2025 - Expires: 6/1/2025
Credential ID: 8VZEYA0XWRH9
Endovascular repair of the thoracic aorta, also referred to as thoracic endovascular aortic repair (TEVAR), refers to a minimally invasive approach that involves placing a stent-graft in the thoracic or thoracoabdominal aorta for the treatment of a variety of thoracic aortic pathologies. In contradiction to open surgery, TEVAR results in reduced recovery times and potentially improved survival rates. Feasibility of TEVAR and correct endograft sizing are based on measurements of Ishimaru’s proximal landing zones. However, TEVAR of the aortic arch still carries a significant risk of medium and long-term complications, including endoleak, endograft migration, and collapse. This may be due to its complex structure and computation of geometric parameters, such as angulation and tortuosity can help to avoid hostile landing zones. The primary goal of this project is to segment the aorta from provided CT scan images, map the landzones (Ishimaru’s proximal landing zones Z0, Z1, Z2, and Z3), and compute various geometric parameters of the aortic arch.
View Project7/1/2023 -3/31/2024
This project presents an automated Computer-Aided Detection (CAD) system for breast mass classification using deep learning techniques. The system is designed to assist radiologists in the early detection of breast cancer by performing three key functions: suspicious region identification, mass/no-mass detection, and benign/malignant mass classification.
View Project12/1/2016 -3/1/2017
Developed an advanced LSTM+CNN model with attention mechanism for Air Quality Index prediction across Indian cities. Our model outperforms state-of-the-art approaches in accuracy and robustness, providing reliable predictions to help mitigate pollution exposure.
View Project1/1/2023 -3/31/2023
Developed a CNN-LSTM hybrid model for detecting AI-generated speech using the ASVspoof2019 dataset. The proposed system achieves state-of-the-art accuracy in classifying audio as 'bonafide' (human) or 'spoof' (AI-generated), combining time-domain analysis with advanced feature extraction. Includes a Streamlit web interface for real-time audio classification.
View Project9/24/2023 -11/30/2023
Verified AI Developer
3-5 years of experience
Preferred commitment: Full Time
Take the next step and bring this top talent to your team
Hire Priya for your team