Apr 2022 – Sep 2022 · 5 mos

• Building ML pipelines and automating ML systems.
• ‣ Built scalable orchestration solution using Airflow pools and Databricks instance pools resulting in reduction of compute costs by more than 2X.

Jul 2019 – Mar 2022 · 2 yrs 8 mos

• Building ML pipelines and automating ML systems.
• ‣ Designed and built a python library to promote reusable, structured feature extraction and model development code. Using FastAPI empowered the delivery team to set up new use cases for the built
• models.
• ‣ Built an end-to-end Alternating Least Squares + Locality Sensitive Hashing based recommendation system using PySpark, MlFlow, Redis to produce recommendation output at item to item level and did performance tuning of the same. Which makes it efficient to provide recommendations based of item clicked in real-time.
• ‣ Using Structure Streaming and Delta Lake built a ML framework to find trending products based on user interactions flowing through Kafka.
• ‣ Using MlFlow, setup model tracking, and model registry workflow to make it effortless to track experiments and deploy specific versions of models.
• ‣ Created model to predict customer lifetime value and next buy probabilities to segment users for better targeting.
• ‣ Created model to extract seasonality component from the user events data of multiple years to bias recommendations to show seasonal products improving customer engagement.
• ‣ Built a tuning framework to tune SOLR boost queries to avoid clumping of similar products together on a category listing page while providing relevant recommendations in real-time.

Apr 2018 – Aug 2018 · 4 mos · India

• Worked on the project “A way to associate listens with MBIDs”. This project aims to create a mapping from unclean data sent by any user, to clean data curated by registered users.
• ‣ Wrote scripts to execute series of SQL queries to perform associations.
• ‣ Tuned SQL queries to increase the efficiency of the queries to perform associations in minutes as compared to hours.
• ‣ Utilized RabbitMQ to push the data across systems and wrote API to fetch the associations for further analysis.

May 2017 – Jul 2017 · 2 mos · Mandi, Himachal Pradesh, India

• Worked on a research project titled “Classification for Breast Cancer Histopathology
• Images”. Machine Learning, Feature Extraction and Stain Normalization are used to classify
• histopathology images into benign (not-cancer) and malignant (cancer).

May 2016 – Jul 2016 · 2 mos · India

• In this internship I worked on creating web applications in Java. I created and deployed a RMI chatting application with multi-threaded server for office use and created backend for Automated Parking system.