“Abhinav and I worked together on diverse projects during his ZS stint as part of Advanced Data sciences track under Business consulting group . Projects ranged form strategy alignment, Research on Real world evidence, POCs for AI capability development, E2E digital transformation and more which could be informed or benefitted with AIML on available data of varied richness. These projects involved optimization for resource planning, forecasting for content modulation, graph networks to inform influencers and marketing leads, recommendation engines for patient oriented care and pure play prediction for early intervention. Abhinav is dependable, resourceful and ramps up quickly in new terrains with little to no guidance . His grasp on fundamentals is strong and his problem formulation skills have improved over the years. He communicates freely, effectively and with poise . It would be good to work with him again and I wish him all the luck for his future . ”
About
Contributions
-
How can you ensure that your machine learning data is realistic?
Often times I’ve seen data from two different sources not communicating similar insights or messaging. This to me becomes a red flag and needs a deep dive on the methodology with which the data has been sourced - what type of sampling bias could it have been exposed to. Once sources of truths are recognized the next step is to check if it passes the general sense check. This could be statistical or from conventional domain knowledge.
-
What are the best techniques for handling imbalanced datasets in deep learning architectures?
Apart from general experiments with under sampling, over sampling, class weights - I try to have a bootstrap mentality by taking in the minority class and sampling equal number of samples from majority class to run one model, and running several such models before reaching an ensemble. The more number of times we’d have random sampling of majority class makes up for the fact that each model is changing the underlying distribution. Ofc works better if models are not too complex.
Activity
-
Amazon is relentlessly focused on earning trust with our customers and selling partners, and we work hard at this every day. One way we are doing…
Amazon is relentlessly focused on earning trust with our customers and selling partners, and we work hard at this every day. One way we are doing…
Liked by Abhinav Sharma
-
Meta will *not* release the multimodal versions of its AI products and models in the EU because of an unpredictable regulatory environment. This…
Meta will *not* release the multimodal versions of its AI products and models in the EU because of an unpredictable regulatory environment. This…
Liked by Abhinav Sharma
-
We just announced a great partnership to reduce environmental impact and waste ! #sustainability #wholefoodsmarket #amazon
We just announced a great partnership to reduce environmental impact and waste ! #sustainability #wholefoodsmarket #amazon
Liked by Abhinav Sharma
Experience & Education
Licenses & Certifications
Projects
-
National Digital Library (NDL)
The project involved data base management using MySql, DSpace Interface, XML programming and reorganizing data for the 100-crore NDL project by government of India.
-
Expedia Hotel Cluster Recommendation System
-
Predicted user preference amidst 100 hotel and property clusters based on users' past travel and stay arrangements. The project necessitated exploring multi-class classification prediction for over 2 million users using Machine Learning Algorithms and intuitive Feature Engineering.
Honors & Awards
-
Analyze This
American Express India
Achieved rank 8th on final leaderboard at the end of inter-IIT online hackathon 'AnalyzeThis' by American Express. Used ensembled ML Algorithms to predict the candidate a voter will choose to vote, given his historical data and specific engineered features, using eXtreme gradient boosting and multinom multi-class classification.
-
Young Data Scientist 2016
ZS Associates
Secured under-30 All India rank, top 1% in the Young Data Scientist competition organized by ZS Associates and got invited to their Pune office. The competition comprised of a Pharmaceutical companys' unsupervised data, which was solved by ensembling statistical inferences with positive learning through SVM one-classification.
-
JEE Advanced (2013)
Indian Institutes of Technology
203/360 score to avail All India Rank of 1,858 out of 150,000 selected candidates from JEE Mains
-
JEE Mains (2013)
National Testing Agency, India
(256/360) to avail an All India Rank of 767 out of 1.5M candidates (99.89 percentile)
-
Maharashtra Talent Search Examination
N Wadia College, Pune Maharashtra
Bagged a state-wise rank 19th amongst 100k+ participating students
Languages
-
English
Full professional proficiency
-
Hindi
Full professional proficiency
-
Marathi
Limited working proficiency
Recommendations received
1 person has recommended Abhinav
Join now to viewMore activity by Abhinav
-
Facebook just released a very different kind of LLMs. These are focused on code optimization and compiler optimization. This is super important…
Facebook just released a very different kind of LLMs. These are focused on code optimization and compiler optimization. This is super important…
Liked by Abhinav Sharma
-
Today, we opened the doors to another great new store in Huntington Station, NY, the 7th store located on Long Island, carrying over 2000 local items…
Today, we opened the doors to another great new store in Huntington Station, NY, the 7th store located on Long Island, carrying over 2000 local items…
Liked by Abhinav Sharma
-
Anthropic just added Artifacts feature to the Claude LLM. This is literally one of the best features I've seen in the LLM space. You can ask it to…
Anthropic just added Artifacts feature to the Claude LLM. This is literally one of the best features I've seen in the LLM space. You can ask it to…
Liked by Abhinav Sharma
Other similar profiles
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore MoreOthers named Abhinav Sharma in United States
-
Abhinav Sharma
-
Abhinav Sharma
Sr. ServiceNow Developer
-
Abhinav Sharma
-
Abhinav Sharma
-
Abhinav Sharma
Talent Acquisition Specialist | Technology Advisor at Stanley Alliance Corporation
158 others named Abhinav Sharma in United States are on LinkedIn
See others named Abhinav Sharma