Career Profile
I have over 6 years of work experience in Data Science / Machine Learning and building production ready tools. My projects span a wide spectrum of applications and techniques: Automated Essay Scoring (NLP) to Robust AI (CV) to Aircraft performance prediction (Stochastic Regression).
My current focus is on Retail Analytics - specifically on targeted promotions and optimal promo design.
Work Experience
- Devised scalable end-to-end recommendation system for consumer retail goods using SynapseML package.
- Achieved 100% improvement in NDCG metric in comparison to vanilla frequency based recommendations.
- Analyzed cross-product interactions to identify halo/cannibal product pairs and enhance promotion planning.
- Developed robust defense against adversarial attacks on ground-based and overhead tracking in PyTorch.
- Improved accuracies of Faster-RCNN based single- and multi-object trackers by 10 and 15 percentage points respectively.
- Designed novel small scale filtering strategy to negate the effect of moving adversarial patch.
- Achieved 4x increase in computational throughput by optimizing function calls and variables within GPU.
- Built stochastic regression toolbox based on Gaussian process regression to fuse multiple data sources and provided quantitative estimates on uncertainty, resulting in significant reduction in aircraft certification costs.
- Supervised the integration of statistical modeling framework from Matlab into Python to achieve seamless interaction with Boeing’s existing design tools, thus achieved multi-fold increase in the toolbox’s reach.
- Formulated cross-functional research projects to meet the requirements of business and engineering units.
- Characterized large-scale coherent structures in swirling jets through dimensionality reduction (PCA, DMD) and established the causality based on the underlying physics through global stability analysis.
Projects
Reports for each of these projects may be obtained by clicking the project name. However, the codes cannot be shared due to Georgia Tech policies. I am listed as "Vamsi Ravilla" or "Vamsi Krishna Chakravarthy Ravilla" in these projects.
Coca-Cola Pricing Strategy [video]
- Effect of price promotions in Spanish beverage market.
Automated Essay Scoring - I [report]
- Single metric essay evaluation using classical supervised and unsupervised learning methods.
Automated Essay Scoring - II [presentation]
- Multi-metric essay evaluation using deep language models and data augmentation using auto-encoders.
Crude Oil Price Prediction [report]
- Is public sentiment on crude oil price related to its hidden state in HMM?
Airbnb Price Prediction [presentation]
- Predict rental prices in New York City using multi-linear regression.
Humana-Mays Healthcare Analytics [report]
- Predict housing insecurity (case study competition)
Publications & Conferences
I am listed as “R. V. K. Chakravarthy” in these articles.
Boeing Technical Excellence Conference, Anaheim, USA, 2019
Journal of Fluid Mechanics, vol. 835, 654 - 673. (2018)
Physics of Fluids, vol. 30(8), 086101. (2018)
16th European Turbulence Conference, Stockholm, Sweden. (2017)
The International Conference on Jets, Wakes and Separated Flows, KTH, Stockholm, Sweden. (2016)
24th International Congress of Theoretical and Applied Mechanics, Montreal, Canada. (2016)
Journal of Fluid Mechanics, vol. 780, 344 - 369. (2015)
EUROMECH Colloquium 567 - Turbulent mixing in stratified flows, Cambridge, UK. (2015)
APS: 67th Annual DFD meeting, San Francisco, USA. (2014)
10th European Fluid Mechanics Conference, TU Denmark, Copenhagen, Denmark. (2014)
Physics of Fluids, vol. 24(10), 102109. (2012)