Research Scholar
I am an MS by Research student in Artificial Intelligence and Data Science with a strong foundation in applied machine learning and deep learning. My work focuses on understanding models beyond surface-level performance, emphasizing objective design, optimization behavior, and evaluation of model reliability.
I have hands-on experience with vision models and ML systems, and a solid theoretical grounding in generative modeling, including loss formulations, KL-divergence, and ELBO-based objectives. I am particularly interested in how optimization principles influence generalization, robustness, and reasoning in modern learning systems, and I aim to transition this understanding toward multimodal and language-based models.
International Institute of Information Technology Bangalore • Full-time
Jul 2025 - Present · Bengaluru, Karnataka, India
Research & Development in Artificial Intelligence & Data Science, specifically focused on Optimization and Deep Generative AI. Optimizing Diffusion AI Model parameters and Optimization in Energy systems.
BigMint (formerly SteelMint/CoalMint) • Internship
Feb 2025 - May 2025 · Raipur, Chhattisgarh, India
Skills: Python, Time Series Analysis, Deep Learning, Data Preprocessing
IIIT-Naya Raipur • Internship
Aug 2023 - Nov 2023 · 4 months
Part of Research & Development under Dr. Santosh Kumar (Professor CSE). Published paper on "UDR Fused Multimodal Approach for Disease Classification in Large Scale Datasets with Advanced CNNs".
Skills: Machine Learning, Deep Learning, Computer Vision, CNNs, Multimodal Learning
IIIT-Naya Raipur • Internship
May 2023 - Jul 2023 · 3 months
Selected as ML Intern under Dr. Anurag Singh in the Outreach Internship Program. Developed IoT-enabled platform using ML algorithms for fitness scoring. Submitted research paper to NCC.
Skills: Machine Learning, IoT, Feature Engineering, Model Development
Conference Paper IEEE • Apr 9, 2025
Authors: Sachin Mishra and Co-authors
Tools & Technologies: LangChain, Chroma DB, RAG Fusion, Vector Embeddings, Python, Large Language Models
This study explores advancements in Retrieval Augmented Generation (RAG) systems tailored for PDF-based question answering. By leveraging domain-specific PDFs as input data, the proposed framework incorporates advanced retrieval techniques, including multi-query generation and RAG fusion, to enhance generated responses' relevance and contextual accuracy. The integration of advanced retrieval strategies significantly improves precision, reduces hallucinations, and enriches response quality for knowledge-intensive tasks.
Conference Paper IEEE • Mar 7, 2025
Authors: Sachin Mishra and Co-authors
Tools & Technologies: Deep Learning, Transfer Learning, MesoNet-4, Conv2D, CNNs, Python, TensorFlow
This research introduces TransDFD, a novel deepfake detection system developed at the mesoscopic level using a deep learning approach and deployed as "Deepfake-guard-AI". The model enhances the original MesoNet-4 architecture by integrating transfer learning with additional layers, optimizing performance on Indian datasets. The system successfully identifies video tampering methods like Deepfake and Face2Face manipulations with high accuracy.
Conference Paper IEEE International Conference on Ambient Intelligence in Health Care (ICAIHC) • Jan 10, 2025
Authors: Sachin Mishra and Co-authors
Tools & Technologies: Convolutional Neural Networks (CNNs), Computer Vision, Deep Learning, Python, TensorFlow, Image Processing
This research introduces a novel, fully automated system for detecting diseases in medicinal plant leaves using Convolutional Neural Networks (CNNs). Addressing the limitations of manual disease identification, which is time-consuming and prone to errors, the proposed system leverages computer vision and deep learning to classify and identify diseases with high accuracy (98.67%), ensuring efficient agricultural health management.
Journal Article Institute of Electrical and Electronics Engineers (IEEE) • Jun 21, 2024
Authors: Sachin Mishra and Co-authors
Tools & Technologies: IoT, Machine Learning, Python, Data Analytics, Health Vitals Monitoring, Wearable Sensors
This paper presents a novel solution featuring an IoT-ML enabled platform for self-monitoring of important health vitals and a fitness assessment score derived through ML algorithms. By integrating data on various health parameters including heart rate, blood oxygen levels, and sleep patterns, advanced machine learning models analyze this data and predict the fitness score that accurately reflects an individual's overall health and fitness level, enabling proactive well-being management.
Book Chapter SPRINGER LNNS (Scopus Indexed) • Dec 23, 2023
Authors: Sachin Mishra and Co-authors
Tools & Technologies: U-Net, DenseNet201, ResNet50, CNNs, Support Vector Machines (SVM), Deep Learning, Python
This paper presents a holistic strategy for breast cancer multiclass disease classification employing advanced deep learning architectures, including U-Net, DenseNet201, and ResNet50 models (referred as UDR fused multimodal). By integrating SVM components to optimize model training on large-scale datasets, an ensembled model was built combining these networks based on their weights and architectures, culminating in a highly accurate classification system with 91.43% accuracy for medical imaging analysis.
Not an experienced one, but here are a few quality projects developed by me, focusing on contextual and functional design.
Have a question or want to collaborate? I'd love to hear from you. Drop me a message and I'll get back to you as soon as possible.
Research Scholar
MS by Research | AI & Optimization | Deep Learning
Get in touch via email or social media
Sachin.Mishra@iiitb.ac.in
