SOURISH SARMAH

Built a serverless architecture to sync mission-critical medical data using AWS Lambda along with API Gateway, and AWS Serverless Application Model(SAM) for infrastructure. Implemented Redis-based caching to optimize the performance of systems accessing frequently-used data and effectively reduced the latency of the system by 25 times. Optimized database design and indexes in PostgreSQL Timescale database, resulting in 60 to 90 times better query performance. Optimized a microservice by migrating its code base from Python to Go which resulted in 20 times faster API responses and better load handling.

Built a microservice for sending notifications to the user interface using socket.io which consumes messages from a message queue and pushes notifications to the user interface using socket connections. Implemented a resilient asynchronous data export feature to export large-scale data from the database to cloud storage in different file formats using multipart upload. Implemented and Integrated a custom ML model training functionality for Loopr through a robust MLOps pipeline using Azure ML Studio, which enabled the platform to streamline the training of labelled data generated on the platform. Led the backend development of a retail application for end-to-end goods ordering and shipment tracking. Built a backend server using Django and also implemented PostgreSQL Full-Text Search. Built data platform and ETL pipelines for TataCliq, one of India’s leading online marketplace and also engineered datasets to power multiple machine learning models using AWS Glue.

Built an ETL data pipeline to process around 1 million raw data points and used Firebase to store these processed data points. Also, implemented an API using Flask to fetch the data from the Firebase with various query filters. Designed and implemented RESTful APIs for a Search Engine efficiency measurement tool. Optimized MongoDB queries and used Celery along with Redis to upload large data files asynchronously.

Worked on CRM Project Development using Flask and PostgreSQL. Fixed issues related to PostgreSQL queries and deployment of the application. Designed and implemented an automated system to identify emails from mailboxes having similar contents using cosine similarity on the TF-IDF matrix of mail content and extract essential links and information.

Relevant courses with Outstanding grades- Introductory Computing, Data Structures, Design and Analysis of Algorithms and Object-Oriented Programming.