Sanskriti Gupta
I am a computer science and design student at Madhav Institute of Technology and Science, Gwalior. My research focuses on speech intelligibility prediction, deep learning, and the space where language meets code. I think of myself as a curious kid who never stopped asking why.
Before my research at IIT Jammu, I developed a machine learning pipeline to detect fraudulent job postings at 3Skill, and built a real-time image captioning application called VisionSense. I have also written technical content for Codeveda and creative pieces for FrameFlicks.
Outside of AI research, I write poetry and short stories, solve data structures and algorithms problems, and I am always looking for the next meaningful challenge.
“The thread that runs through everything I do is curiosity. I ask why until I reach the bottom of it. That question is what drove me from competitive programming to deep learning research to writing poetry. All of it is the same thing: trying to understand.”
Work & Research
Developed a non-intrusive speech intelligibility prediction framework using self-supervised speech representations from Whisper and Wavelet Scattering Transform features.
Engineered acoustic feature extraction pipelines with Gammatone filterbanks, Whisper embeddings, and PyTorch-based deep learning models for intelligibility assessment.
Achieved Development RMSE of 21.62, surpassing all reproduced baselines on the evaluation dataset.
Built a machine learning pipeline to detect fraudulent job postings using structured job attributes and textual features extracted from job descriptions and company profiles.
Performed data preprocessing, text cleaning, feature engineering, and TF-IDF vectorization across multiple classification models including Logistic Regression and Naive Bayes.
Wrote technical content on AI and ML topics, including Large Language Models and their real-world applications.
Write poetry, humor pieces, and short stories. FrameFlicks is a creative outlet that keeps my writing sharp and my thinking flexible.
Education
SGPA: 8.55
91.6%
95.8% (Top 3 in district)