Dec 2024 – Present · 1 yr 3 mos · San Francisco Bay Area

Nov 2024 – Present · 1 yr 4 mos

Nov 2024 – Present · 1 yr 4 mos

Jan 2024 – Present · 2 yrs 2 mos · San Francisco Bay Area

Oct 2023 – Dec 2024 · 1 yr 2 mos · San Francisco Bay Area

• Led the Y Combinator–Microsoft partnership as an AI advisor, helping startups build and scale AI/ML workloads on Azure, covering fine-tuning, RAG, PromptFlow, and MLOps best practices.
• Delivered advanced GenAI workshops and office hours to support startups with distributed training and inference on H100 clusters.
• Curated best-practice guides on model selection, fine-tuning, RAG/RAFT architectures, scaling with GPU clusters, and implementing secure, bias-aware, and RLHF-aligned pipelines.
• Spoke at keynotes, panels, and conferences on building Responsible AI systems using Azure and open-source tools.

Jul 2023 – Present · 2 yrs 8 mos

Jun 2023 – Present · 2 yrs 9 mos

Jul 2022 – Dec 2022 · 5 mos

Apr 2022 – Dec 2022 · 8 mos

Jan 2022 – Sep 2023 · 1 yr 8 mos

• Working on building solutions using Large Language Models and Generative AI capabilities on Google Cloud- GenAI Specialist
• Customer-facing Data Scientist, building machine learning solutions for customer use cases leveraging core Google Cloud AI platform and services.
• Collaborating cross functionally with product team to perform industry research on machine learning use-cases and
• building technical assets for end-to-end pipelines
• Co-leading initiatives for building tools for Responsible AI workflows
• Led a team to develop Data Quality Assessment Framework to be widely used in Google Cloud AI teams
• Working on demand forecasting models, MLOps pipelines, personalization models, DocAI, and VisionAI models, across retail, banking and manufacturing industries (confidential customers)
• 20% – Working as a PM with the product team to develop features in Vertex AI platform (confidential)

Jul 2021 – Dec 2022 · 1 yr 5 mos

Jul 2021 – Dec 2022 · 1 yr 5 mos

Apr 2021 – Jan 2022 · 9 mos

• Building Data and AI capabilities of Cloud Pak for Data, IBM Lead with offerings and partnerships by building data science and machine learning models to cater to the broader client, sales, and partner communities.
• Worked cross functionally with product team, data science team and sales to research AI use-cases for clients by conducting discovery workshops and building demo assets to showcase the value of IBM Cloud services
• Lead a team to generate new opportunities by proposing new features for IBM Cloud Data & AI product offerings based on industry research and fostering broader client, and product analytics
• Leading post-signature activities for CSM, to ensure client projects are begun quickly and seamlessly.

Feb 2019 – Apr 2021 · 2 yrs 2 mos

• Working on 6-8 week client engagements to implement SoA (state of the art) machine learning models, and integrate them into their organizational workflow to put AI into action.
• Led engagements with clients from multiple industries like health, finance, intergovernmental organizations (United Nations), non-profit organizations, international government bodies.
• Managed cross-border client accounts and traveled to London, Dubai, Istanbul, Mumbai, and multiple states within the United States.
• Submitted a patent for building a novel reinforcement learning model for machine trading.

Apr 2021 – Mar 2023 · 1 yr 11 mos

Mar 2021 – Present · 5 yrs · California, United States

• First of its kind volunteer mentorship platform for folks in Artificial Intelligence!
• Illuminate - AI is a platform built to connect people from the Artificial Intelligence Community.We believe knowledge grows by sharing. We are driven to connect mentors, who would like to give to the community, to mentees, who are seeking guidance and motivation. The mentorship program is not limited to a senior individual mentoring a novice but is it also open to peer-mentorship.
• There is a lot to learn!
• A platform is also for sharing thoughts and ideas in the space of Artificial Intelligence and its applications in real-world use-cases.
• Playlists available are Weekend Musings, AI Guild, What's Tuning and Cracking Data Science Interviews.

Oct 2020 – Jul 2021 · 9 mos · New York, New York, United States

Sep 2018 – Dec 2018 · 3 mos · Greater New York City Area

• Capstone project via Columbia University (With Jared Peterson @ Goldman Sachs)
• Developing a model for finding the Similarity Indexes of various Medical Institutes based on their clinical trial data using the open source AACT Database. (Text analytics and exploring alternative data)

Jun 2018 – Dec 2018 · 6 mos · Greater New York City Area

• Developing Reinforcement Learning Model for Long & Short Trading and Portfolio Management.

Jun 2018 – Dec 2018 · 6 mos · Greater New York City Area

• Working on patient's EHR and time series data of heart rate, blood pressure and patient's health response during the operation. The goal is to develop a forecasting model to detect the seizure of patients; using multi-level model. The time series data would be analyzed using dynamic time warping methods due to varying lengths and unequal intervals of the data.

Apr 2018 – Dec 2018 · 8 mos

Jan 2018 – May 2018 · 4 mos

• Analysis and extension of scikit- learn ( Guide : Prof. Andreas Mueller)
• Scikit-learn is a Python machine learning library containing a large collection of machine learning models, as well as evaluation metrics and tools for implementing machine learning workflows.
• The goal of this project is to analyze the current usage of scikit-learn on a large scale (i.e. the scale of all open-source code, even all public code), and extend the library based on the findings; to identify usage patterns, problematic use cases, and ways to improve the interface.

Sep 2017 – Jan 2018 · 4 mos · Greater New York City Area

May 2017 – Jun 2017 · 1 mo · New Delhi Area, India

• During my internship as a Data Scientist I accomplished the following tasks:
• Data collection and extraction using various websites like World Bank, UNDP, UNICEF etc.
• Data filtering
• Data visualizations using D3js predominantly along with Microsoft Power BI, Tableau and Adobe Photoshop for Info graphics.
• Writing tech blog correlating topics like GST, BRICS Meet, Ocean Conference, Planet 50-50 with SDGs.
• Extensive reading and research on data and whitepapers in SDG 13 and 14: Climate Change and Life below water respectively.
• Whilst my term, we worked on the project of 'The Ocean Conference Challenge' as well, for data analytics and visualization, for which we were awarded by the UN-DESA.

Dec 2016 – May 2017 · 5 mos · New Delhi, Delhi, India

May 2016 – Jul 2016 · 2 mos · Hyderabad Area, India

• Worked with the Bing Team for Machine Learning and NLP.
• Worked on developing a Skype ChatBot for Redmond Campus, fulfilling 6 broad criteria.
• Coordinated with the Redmond team for understanding the functional and technical requirements.
• Worked with BotFramework in Microsoft Visual Studio (C#) on Azure, Cognitive services, Bing Maps, NLP, luis.ai, REST APIs and machine learning.

Jan 2016 – Feb 2016 · 1 mo · New Delhi Area, India

• Did extensive research on capacity utilization of goods and Index of Industrial Production.
• Read many research papers and company white papers to gain knowledge.
• Made theoretical hypothesis regarding the IIP growth, and to prove the hypothesis, did data analytics.
• The analytics included Time Series and Wharton Index of Growth for Capacity Utilization of Goods in Indian Industry using Index of Industrial Production.

Dec 2015 – Feb 2016 · 2 mos · Delhi

• Data set of a restaurant-customer ratings was collected through internet sources. (data.gov.in)
• The analysis had to be done to find the dependent factors which influence the overall rating, food rating and service rating of the restaurant.
• Predictive analysis needs to be applied to find response of customer in a different scenario.
• In the pre-processing of the dataset, qualitative information was converted to quantitative data; including data cleaning.The dataset was converted into multiple table using the database approach- entity relationship diagram.
• Supervised learning methods of correlation, linear regression and ANOVA was done. Later, unsupervised learning method was also applied. Unsupervised learning does not use previously known results to train its models. Rather, it uses descriptive statistics to examine the natural patterns and relationships that occur within the data and does not predict a target value.
• Another data set of Mutual Fund Performance for 8 years was collected.
• The growth rates of the mutual fund are influenced by various factors, which are mostly indirect, like the technological development in industries, industrial growth, gross profit margin in the industry etc. in which the mutual fund company has invested its bonds. For the customer to buy a mutual fund bond, the Net Value of the bond should be high, so as the person who is investing gets higher returns.
• For the purpose, predictive and time series analysis was applied for model generation.

Jun 2015 – Jul 2015 · 1 mo · New Delhi Area, India

• I did the project with Dr. Neena Sondhi, who has a diverse 20+ years of diverse experience of teaching, research and consultancy. An alumnus and a merit holder from the Faculty of Arts, University of Delhi, she has successfully managed senior administrative roles in the organization.
• The project was co-mentored by Dr. Chhavi Mehta who is a professor in IMI. She has completed her PhD from IIT, Delhi and has done Empirical Financial and Accounting Research Program from IIM, Calcutta.
• A survey was conducted among various Stock market investors to analyse their investment behavior.
• The data collected was converted from qualitative to quantitative for ease in analysis. Data cleaning and pre-processing was performed.
• The aim of the project was to analyse and build models for Customer Satisfaction and Profit Prediction & Enhancement for Stock Market.
• The analysis was done using IBM SPSS.

Mar 2015 – May 2015 · 2 mos · Chennai

• Data from NSE and BSE websites of Sensex and Nifty was collected.
• Data values of various parameters like gold rates, industrial production, textile rates, mining etc were extracted.
• The purpose was to find a relationship, as to how the growth or decline in various industries can effect the Sensex and Nifty rates to go up or down.
• Predictive modeling using regression and time series analysis was applied.
• The analysis was done in IBM SPSS.

Dec 2014 – Jan 2015 · 1 mo · New Delhi Area, India

• I worked for the client DirecTV, USA for modeling their transmission via cloud platform.
• Extensive research was done to understand the encryption algorithms, video streaming, real time transmission using cloud computing etc.
• A cloud architecture was designed to do receive data from satellite, perform video encryption and transmit it in real time to the receiver set top boxes.

Jan 2015 – Aug 2015 · 7 mos · Virtual

• The company hired me for social media and business analytics- highly confidential.
• The dataset was entirely generated by extracting information from the Twitter profile of the clients for whom the analytics was performed. For the purpose, Twitter Streaming API (text mining) was used.
• The data was wired to the RStudio using twitteR library and the analytics was coded using R language.
• Various analytical methods were applied, namely, word cloud where an image composed of words used in a particular text or subject, in which the size of each word indicates its frequency or importance; sentiment analysis which is a process of computationally identifying and categorizing opinions expressed in a piece of text, especially in order to determine whether the writer's attitude towards a particular topic, product, etc. is positive, negative, or neutral.
• Later User Activity Monitoring (UAM) was also performed to visualize the activeness of the user in making posts, sharing posts, liking and commenting using info graphics.
• All the data inferences were converted into one-sliders which consisted of all information as graphics for each client being inspected.

Jun 2014 – Jul 2014 · 1 mo · New Delhi Area, India

• Extensive research on forensic science and handwriting analysis was done.
• Handwriting features and their dependency with the human personality was studied.
• A survey was conducted where uniform handwriting samples of families were collected, father, mother, grandparents, kids and siblings; in order to analyse upon the characteristics of handwriting that are inherited.
• Later, a software was developed using MATLAB, where handwriting text can be recognized, similar to OCR and handwriting features like slant, kurtosis, skewness, margin gap, inter-word and inter-line spaces were analysed.
• Our research and the results were well-appreciated by a lot of researchers, and we got our work published in International Refereed Journal of Engineering and Science.