Resume

- Lead backend development using .NET and C#, building scalable services for a real‑time safety monitoring platform.
- Integrate AI/ML features to improve anomaly detection accuracy and enhance platform intelligence.
- Develop automated testing frameworks and CI workflows to increase reliability and reduce manual QA effort.
- Collaborate closely with IoT hardware teams to ensure seamless communication between sensors, fire alarms, and backend systems.
- Build automation tools in Python and C# to streamline internal workflows and improve operational efficiency.

- Designed and automated ETL workflows using Alteryx and Python, improving data accuracy and reducing manual processing time.
- Built NLP classification models that reduced unnecessary truck dispatches by 45%.
- Developed churn and satisfaction prediction models, contributing to a 24% reduction in customer complaints.
- Created pipelines to ingest geospatial and operational data into the enterprise data warehouse.
- Built an ETL pipeline for Ookla SpeedTest data, reducing search and retrieval time by 40%.

- Migrated a 5,000+ product catalog from Drupal 8 to WordPress.
- Built JSON API integrations to compare competitor pricing in real time.
- Optimized SQL queries and backend logic, improving system response by 35%.

- Developed OCEL.AI, an end‑to‑end ML tool used by 4 universities and 100+ active research projects.
- Built visualizations for structured, unstructured, text, and image data using D3.js, T‑SNE, and Tableau.

Grants: Dean’s International Student Award (DISA), Chandra Scholarship
Coursework: Design Analysis of Algorithms, Deep Learning, Cloud Computing, Web Programming, Big Data Analytics, ADBMS, Unity AR

Location: Dehradun, India

See Publication
We proposed a generic framework for diagnosis of Progeria syndrome among the newborns. We proposed novel framework architecture of VGG-16. We have used multiple machine learning algorithms and feature extraction tools and we got best results in Logistic Regression algorithm.

See Publication
We propose a generic model to classify a given image by detecting a specific image patch. Employing MPEG-7 features along with feature selection populates the feature space which is used to train using SVM classier.

Developed FeedKC website to connect food donors and requestors online. This project had reduced food waste by more than 700 lbs per semester. That's 700+ lbs of food directly translating into dollars saved for their bottom line in just 4 months.
Website: www.feed-kc.com