ABHAY YADAV

Experience

• 

  Oracle

  Jul 2009 - Jul 2014

  Primary Role: Full responsibility for software project lifecycle, from conception to deployment.Leadership and Development: Dedicated to the mentorship and skill enhancement of new team members, including new hires, interns, and contractors, through hands-on guidance and training.Project Involvement: Actively engaged in project design and coding as an individual contributor.Product Insight: Worked on OWSM, a Java/J2EE-based security and monitoring solution for Web services, incorporating technologies such as EJB and MBeans. Show less

  • Senior Member of Technical Staff

    Sept 2012 - Jul 2014

  • Member of Technical Staff

    Jul 2009 - Sept 2012
• 

  University of Maryland – College of Computer, Mathematical, and Natural Sciences

  Jun 2015 - Dec 2021

  Machine Learning Research Assistant

  Conducted extensive research, development, and implementation of cutting-edge deep learning methods for Computer Vision applications. Successfully shipped numerous innovative solutions to various stakeholders, including funding agencies, leading to the publication of several research papers in prestigious conferences such as ECCV, NIPS, ICLR, AISTATS, and BMVC.

• 

  Intel Labs

  Jun 2018 - Dec 2018

  Machine Learning Intern

  Researched and developed advanced second order methods for deep neural network training, significantly enhancing model efficiency. Implemented and shipped innovative solutions, contributing to a published paper in the field.

• 

  Comcast

  Jun 2021 - Aug 2021

  Machine Learning Intern

  I researched, developed, and implemented an innovative approach in our in-house recommender system by incorporating knowledge graph embeddings of movie metadata. This development was pivotal in the successful launch of a new product feature, Dynamic Menu Generation, utilizing multi-hop inference for enhanced performance. This advancement notably improved program coverage by around 19%, benefiting our extensive customer base of 27 million.

• 

  Comcast

  Dec 2021 - now

  Senior Machine Learning Researcher (Applied Scientist - AI/ML)

  Position: Senior Machine Learning Researcher at Comcast.Interest Areas: Focused on tackling large-scale machine learning challenges, specifically in the fields of recommender systems, generative AI, knowledge graph embeddings, conversational AI, computer vision, and optimization techniques.Current Focus: Engaged in fine-tuning generative AI models for enhanced efficiency, integrating these with knowledge graph embeddings to advance recommendation systems. We do this through custom LLM models pretraining, LoRA finetuning, RAG based retrieval augmentation.Key Projects: 1) Developed (from scratch) and deployed in production a Learning To Rank (LTR) models leveraging contrastive learning for slate recommendation, achieving a 1.5 % performance increase (over ~25 M customers) verified through A/B testing. 2) Developed and deployed in production a positional debiasing Learning To Rank(LTR) model for slate recommendation, achieving a 1% performance increase(over ~25 M customers) verified through A/B testing.Innovative Research: Pioneering a multi-objective approach for slate recommendation using Reinforcement Learning from Human Feedback (RLHF, DPO), currently in A/B testing phase. Contributed two new metrics for offline A/B test evaluation (used in production before launching a new AB test): RIPS and SNIPS.Proof of Concept: Executed a project using LangChain to create an interactive movie chatbot, capable of playing movies or taking actions based on user conversations. Show less