Rucha Khopkar
ML Research Scientist @ Cognitiv | Carnegie Mellon | Machine Learning
- Role
- Ml Research Scientist at Cognitiv
- Location
- Bellevue, WA, US
- LinkedIn followers
- 500 followers
About Rucha Khopkar
As a dynamic data scientist with a proven track record spanning 1.5+ years in the industry and a comprehensive 4+ years of professional immersion in machine learning and data science, my commitment lies in delivering tangible impact through the fusion of cutting-edge technologies and insightful analytics. My overarching goal is to empower organizations to make informed, data-driven decisions and fostering efficiency gains.Technical Prowess:I possess proficiency in a diverse array of tools, including Python, R, SQL, Tableau, and a wide spectrum of machine learning and deep learning techniques. My expertise extends to specialized frameworks such as PyTorch, TensorFlow, and HuggingFace, where I apply advanced modeling techniques like generative AI, natural language processing, and computer vision to extract meaningful insights from complex data landscapes.Contact Information : r••••••••@gmail.com
Experience
Ml Research Scientist
May 2024 — Present · Bellevue, WA, US
Refactored PyTorch code to train multi-class and regression models in PyTorch Lightning, enabling a smoother migration to the new framework. Optimized training by improving loss calculation and enforcing a minimum of 5 epochs before checkpointing and early stopping. Ensured the updated models maintained 99% performance parity with the previous framework. The new code is now used by 2 out of 18 team members, improving model scalability and ease of training. Additionally, created Docker images to simplify execution and deployment.2. Developed a Video Completion Rate (VCR) Autotuner for the bidding system, leveraging a PID controller to maximize VCR performance. This initiative enabled 13 new advertisers to optimize campaigns where VCR was the key performance metric.3. Optimized and maintained machine learning models to drive KPI performance for Progressive, our largest advertiser, achieving a +10% lift in CTV campaign performance and meeting 100% of margin and incremental CPA targets. Led model refreshes, feature enhancements, and data source optimizations to improve bidding efficiency.4. Increased model AUC by 1.2% by experimenting with prevalence rate adjustments and implementing a zerochecklayer, which validated input completeness and improved performance.5. Enhanced data selection strategy, initially shifting from general campaign data to bidding-aligned datasets, improving CTV performance by 6%. Later transitioned to fully CTV-based training, further boosting KPI achievement by 10%.6. Conducted real-time production model analysis, identifying and mitigating irrelevant domain bidding, leading to a 5% reduction in wasted ad spend through domain-based filtering.7. Researched and implemented insights from academic papers, refining bidding strategies and improving model scalability and efficiency.
Education
Pune Institute of Computer Technology
Bachelor of Engineering - BE, Electrical, Electronics and Communications Engineering
Carnegie Mellon University
Master of Science - MS, Electrical and Computer Engineering Applied Advanced Program
Find verified contacts for anyone on LinkedIn
Unifers gives sales teams verified emails and direct dials, enriched profiles, and outreach that lands in the inbox.
Free plan included · No credit card required
This profile is compiled from publicly available professional sources. Unifers is not affiliated with or endorsed by LinkedIn. Request removal of this profile.