Roberts Center for Pediatric Research
734 Schuylkill Ave
Philadelphia, PA 19146
News
Latest Updates
Micro-DualNet accepted to IEEE FG 2026 in Japan! 🇯🇵🌸
Poster accepted to INSAR 2026 in Prague! 🇨🇿
Joined Center for Autism Research at CHOP as Data Scientist I!
Joined Chima Inc. as AI Engineer in San Francisco 👨🏻💻
Graduated with MSE in Data Science from University of Pennsylvania! 
🥳
Presented Paper at 12th ICCNT, IIT Kharagpur 🥳
Education
Academic Background
MS in Data Science GPA: 3.8 / 4.0
University of Pennsylvania
- Big Data Analytics
- Machine Learning
- Deep Learning
- Natural Language Processing
- Computer Vision
- Statistical Learning
- Data Engineering
- Databases & Information Systems
B.Tech in Electronics & Telecommunication GPA: 3.8 / 4.0
Sardar Patel Institute of Technology, Mumbai
Experience
Where I’ve Worked
Children's Hospital of Philadelphia — Center for Autism Research
Developing computational frameworks for quantifying social behavior in pediatric populations through video-based body and facial analysis. Building end-to-end pipelines for 2D-to-3D pose lifting, micro-expression recognition, and behavioral feature extraction on high-performance computing infrastructure. Focused on bridging machine learning research with clinical workflows to support autism diagnostics and intervention studies.
Chima Inc., San Francisco
Designed and deployed multi-agent LLM pipelines for enterprise applications. Managed Kubernetes-based infrastructure for scalable model serving and orchestration.
The Wharton School, University of Pennsylvania
Conducted large-scale ESG analytics over 1M+ news articles using AWS services (S3, Athena, Redshift). Built automated data pipelines and dashboards to support research in sustainable finance.
Publications
Selected Research
Micro-DualNet: Dual-Path Spatio-Temporal Network for Micro-Action Recognition
Facial Expression Asymmetry as a Novel Nonverbal Marker of Autism in Naturalistic Social Interaction
Bitbox: Behavioral Imaging Toolbox for Computational Analysis of Behavior from Videos
Identification of Multiple Copy-Move Attacks in Digital Images using FFT and CNN
Pneumonia Detection from Chest X-ray using Transfer Learning Techniques
Projects
Research Projects
Disaster Tweet Classification with Multimodal Learning
Built a multimodal system fusing BERT text embeddings with CLIP/BLIP visual features and attention mechanisms to classify disaster-related tweets. Achieved 79% F1 on a benchmark dataset, demonstrating effective cross-modal information fusion for crisis informatics.
Depression Likelihood Prediction on Reddit
Developed a retrieval-augmented prediction system combining LSTM networks with MentalBERT to assess depression likelihood from Reddit posts. Leveraged domain-specific language models and retrieval-augmented inference to improve sensitivity on mental health discourse.
Skills
Technical Toolkit
Languages & Tools
ML & Frameworks
Data & Infrastructure
Research Areas