Jan 2022 – Jul 2023 · 1 yr 6 mos · Mumbai, Maharashtra, India

• ♦ 𝓔𝓷𝓰𝓪𝓰𝓮𝓶𝓮𝓷𝓽 1 -
• 𝐏𝐫𝐨𝐛𝐥𝐞𝐦 𝐒𝐭𝐚𝐭𝐚𝐦𝐞𝐧𝐭 -> To forecast the profitable market price based on several features
• 🔹 𝗦𝗼𝗹𝘂𝘁𝗶𝗼𝗻 -> Used MLOPs concepts to create a production grade end 2 end pipeline
• 1. Created the architecture that constitutes following components in the Kubeflow pipeline.
• a. Using GCP Dataflow, created the pipeline to migrate the data from On-Premises to GCP Big Query.
• b. Created the following components in end2end Kubeflow MLOps pipeline :-
• ☝𝐃𝐚𝐭𝐚 𝐈𝐧𝐠𝐞𝐬𝐭𝐢𝐨𝐧 -> To ingest the data from the Big Query
• ☝𝐃𝐚𝐭𝐚 𝐏𝐫𝐞𝐩𝐫𝐨𝐜𝐞𝐬𝐬𝐢𝐧𝐠 -> To preprocess and transform the data to formats required for Exploratory Data Analysis and forecasting model development.
• ☝𝐃𝐚𝐭𝐚 𝐒𝐩𝐥𝐢𝐭 -> To split the data into the train, test and valid.
• ☝𝐓𝐫𝐚𝐢𝐧𝐢𝐧𝐠 -> To start the training using the AutoML/ custom training.
• ☝𝐌𝐨𝐝𝐞𝐥 𝐄𝐯𝐚𝐥𝐮𝐚𝐭𝐢𝐨𝐧 -> To compare the model metrics of the previously deployed champion model and recently trained model and write them onto the Big Query table.
• ☝𝐄𝐧𝐝𝐩𝐨𝐢𝐧𝐭 -> Create the endpoint if its already not there.
• ☝𝐃𝐞𝐩𝐥𝐨𝐲 𝐌𝐨𝐝𝐞𝐥 -> Conditional component, which will deploy the model as per the above result on the endpoint.
• ☝𝐃𝐫𝐢𝐟𝐭 𝐄𝐯𝐚𝐥𝐮𝐚𝐭𝐢𝐨𝐧 -> Check the data and concept drift and accordingly send the monitoring alert to the registered email.
• 2. Deployed the whole pipeline and schedule its re-run using Cloud Scheduler and Cloud Function.
• 🔹𝗦𝗲𝗿𝘃𝗶𝗰𝗲𝘀 𝗨𝘀𝗲𝗱 -> 𝘿𝙖𝙩𝙖𝙛𝙡𝙤𝙬, 𝘽𝙞𝙜𝙌𝙪𝙚𝙧𝙮, 𝙑𝙚𝙧𝙩𝙚𝙭 𝘼𝙄, 𝘾𝙡𝙤𝙪𝙙𝙎𝙘𝙝𝙚𝙙𝙪𝙡𝙚𝙧, 𝘾𝙡𝙤𝙪𝙙 𝙁𝙪𝙣𝙘𝙩𝙞𝙤𝙣, 𝘼𝙪𝙩𝙤𝙈𝙇, 𝘾𝙡𝙤𝙪𝙙 𝙋𝙪𝙗/𝙎𝙪𝙗, 𝘼𝙥𝙥 𝙀𝙣𝙜𝙞𝙣𝙚, 𝘾𝙤𝙢𝙥𝙪𝙩𝙚 𝙀𝙣𝙜𝙞𝙣𝙚

May 2021 – Jan 2022 · 8 mos · Toronto, Ontario, Canada

• Personality Detector: Used NLP and classification-based models like SVM and Random Forest to generate BIG-5 OCEAN personality profiles of individuals.
• Automation: Developed a Python script using NLP techniques to identify target sales leads for the company based on various demographic and psychological features, and automated the process of sending connection requests on LinkedIn.

Apr 2019 – Mar 2021 · 1 yr 11 mos

• 🔹 𝐑𝐞𝐚𝐥 𝐓𝐢𝐦𝐞 𝐂𝐨𝐧𝐬𝐮𝐦𝐞𝐫 𝐀𝐧𝐚𝐥𝐲𝐭𝐢𝐜𝐬 : Tracked real time consumer journey on android/ ios app using events and built propensity algorithm to take action and decreased customer churn by 5%. This model optimised funnels across both traditional and digital channels.
• 🔹 𝐂𝐨𝐧𝐬𝐮𝐦𝐞𝐫 𝐂𝐥𝐮𝐬𝐭𝐞𝐫𝐢𝐧𝐠 : Performed customer segmentations by clustering consumer into personalised buckets leveraging RFM (Recency, Frequency and Monetary) and other machine learning algorithms to improve customer retention by 7% month over month. It helped in offering personalised promotional campaigns, loyalty programs and special flyers to customers.
• 🔹 𝐑𝐞𝐜𝐨𝐦𝐦𝐞𝐧𝐝𝐚𝐭𝐢𝐨𝐧 𝐄𝐧𝐠𝐢𝐧𝐞 : Developed Recommendation Engine for the company’s App by using hybrid content collaborative filtering which increased user engagement by 32% from home page to order page.

Apr 2018 – Mar 2019 · 11 mos

• 🔹 𝐈𝐧𝐭𝐞𝐥𝐥𝐢𝐠𝐞𝐧𝐭 𝐂𝐨𝐮𝐩𝐨𝐧 𝐀𝐧𝐚𝐥𝐲𝐳𝐞𝐫 : Designed scalable, secure real time analytical campaign engine using tools like Spark, Kafka, Hive, MongoDB, Python along with various machine learning algorithms and deployed them using FLASK Rest API on production which increased customer engagement by 25% from cart page to payment page. It helped in investigating ROI of marketing initiatives.
• 🔹 𝐂𝐨𝐦𝐦𝐮𝐧𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐒𝐤𝐢𝐥𝐥𝐬 : Collaborated with leaders, to provide data driven insights of marketing performance, customer segments behaviours using visualisation and presentation.
• 🔹 𝐅𝐞𝐞𝐝𝐛𝐚𝐜𝐤 𝐂𝐥𝐚𝐬𝐬𝐢𝐟𝐢𝐜𝐚𝐭𝐢𝐨𝐧 : Demystified Natural Language Processing and other machine learning algorithms to build automatic feedback classifier tool which used to classify approximately 50,000 customer comments with an accuracy of 90%.

Jul 2017 – Apr 2018 · 9 mos

• 🔹 𝐀𝐮𝐭𝐨𝐦𝐚𝐭𝐢𝐨𝐧 𝐨𝐟 𝐕𝐢𝐬𝐮𝐚𝐥𝐢𝐳𝐚𝐭𝐢𝐨𝐧 𝐑𝐞𝐩𝐨𝐫𝐭 : Designed and automated more than 15 innovative reports and dashboards based on key performing metrics for different management levels, leading to increased transparency and sales performance by analysing root cause of changes.
• 🔹 𝐌𝐨𝐝𝐞𝐥 𝐛𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐟𝐫𝐨𝐦 𝐒𝐜𝐫𝐚𝐭𝐜𝐡 : Used classification and regression modelling techniques to build an analytical model to predict optimal price and best confidence interval of the product which increased the contract win probability by 70%. Further deployed it on production using python web service in FLASK API.
• 🔹 𝐂𝐨𝐦𝐦𝐮𝐧𝐢𝐜𝐚𝐭𝐢𝐨𝐧 𝐒𝐤𝐢𝐥𝐥𝐬 : Communicated data driven recommendations to key stakeholders using quantitative analysis, strong data visualizations and superior presentation skills which helped in identifying problems and root cause behind it.
• 🔹 𝐋𝐞𝐚𝐝𝐞𝐫𝐬𝐡𝐢𝐩 𝐒𝐤𝐢𝐥𝐥𝐬 : Coached, developed and motivated team members, by offering them knowledge related to data mining techniques.

Feb 2014 – Jan 2017 · 2 yrs 11 mos

• 🔹 Collected data from data sources like Amazon's Redshift database and used to develop dashboards which was used for the internal partners.
• 🔹 Used cluster analysis and other unsupervised learning algorithms to analyse and solve complex business problems.
• 🔹 Used to discover patterns hidden in large data sets and worked with stakeholders to improve business outcomes.