Raj
Bhadani
ML/AI Researcher working on real-time deepfake detection under social media compression with Dr. Nidhi Gupta at NIT Kurukshetra. Architect of SATYA-ASTRA, submitted to iDEX Open Challenge (₹1.5 Cr). 11 projects across CV, NLP, time-series, network analysis, and defence AI. Led a 30-member university tech society as President.
Current Work
Four-layer pipeline for defence-grade deepfake intelligence: ingest incoming media, detect manipulation with multi-modal ML, trace attribution to source actors, and trigger a coordinated response. Multilingual across all 22 scheduled Indian languages via Bhashini API. Submitted to iDEX Open Challenge (₹1.5 Crore). Research conducted under Dr. Nidhi Gupta, NIT Kurukshetra, focused on deepfake detection robustness under social media compression artifacts (JPEG, H.264).
Technical Stack
Skills & Expertise
Experience
- Conducting research on real-time deepfake detection under social media compression (JPEG, H.264 artifacts)
- Architecting SATYA-ASTRA: a four-layer military deepfake detection & attribution platform submitted to iDEX Open Challenge (₹1.5 Cr)
- Integrating Bhashini API for multilingual support across all 22 scheduled Indian languages
- Completed IBM SkillsBuild AI/ML program; applied green skills and AI concepts to real-world projects
- Built ML models focused on sustainable technology under the Shell-Edunet AICTE framework
- Worked on data preprocessing, model evaluation, and reporting of results
- Completed RISE Internship Program with practical training in data analysis and visualization
- Worked on datasets involving customer and sales analytics
Key Projects
Applied K-Means clustering on customer transaction data after preprocessing and feature scaling to identify distinct customer groups.
Performed exploratory data analysis, trend analysis, and built visual dashboards on e-commerce sales data.
Built a regression model to analyze historical sales patterns and generate future forecasts.
Cleaned and analyzed academic datasets, then built an interactive analytics dashboard.
Built predictive models on historical environmental datasets to forecast future greenhouse gas emission trends.
Performed time-series analysis and predictive modeling on global COVID datasets to analyze pandemic trends.
Applied Social Network Analysis metrics — centrality and connectivity — to study relationships in public health and security systems.
Built classification models on patient health indicators to identify high-risk individuals for early diabetes detection.
Mapped and analyzed a 61-organization Indian terrorist network using social network analysis. Computed betweenness centrality to identify critical nodes and chokepoints in the network.
Fine-tuned ResNet18 on the FER-2013 dataset for 7-class facial expression classification. Outperformed a baseline CNN by 9 percentage points.
Built a Tesla stock price forecasting model using Facebook Prophet. Benchmarked against ARIMA baseline across MAE and RMSE metrics.
Extracurriculars
Led a 30-member team; organized SYNTAX'24 annual tech fest covering logistics, coordination, and execution.
Managed event operations and cross-team collaboration for society technical events.
Oversaw IT operations and digitization initiatives for the rural development program.
Education & Certifications
- Introduction to Data Science
- Fundamental AI Concepts
- Python with Advanced AI
- Introduction to Generative AI
- Introduction to Responsible AI
Let's Connect
Open to internships, collaborations, and AI/ML projects.
rajbhadani9897@gmail.com