Professional Experience

Deloitte US - India

09/2021 – 07/2024   Bangalore
Consultant

Insurance Services Client

• Architected and deployed a scalable Python framework and data pipelines to process, extract, and audit 1800+ attributes for 4 data products with 8M+ records per load cycle.
• Revamped CDC features and triaged 40+ issues in the data warehouse encompassing 7 source systems across 3 LOBs in Postgres and AWS Redshift ensuring data consistency across the platforms with daily loads.
• Designed Data Mesh architecture using Delta Lake and Delta Share framework to connect multiple client LOBs, facilitating cross-functional data access for downstream modeling systems.
• Improved team efficiency by 30% by building config generation utility for Stonebranch workflow system.
• Received 3 Applause awards for developing and delivering Actuarial and ServiceNow systems.

AIOPS.D

• Developed containerized models for Natural Language Query Translation using fine-tuned Transformer architectures and integrated them via microservices and service discovery server on the ServiceNow platform.
• Developed frontend web components to facilitate navigation and user interaction with 3rd party chatbots on Now Experience Framework.

Freelance Developer

01/2021 – 06/2021   Remote
Backend Developer

• Created an order processing service and geography-based automated delivery agent allocation system for a grocery delivery startup, streamlining order tagging process and reducing manual efforts by ∼ 30%.
• Developed Telegram chatbot delivering automated news updates tailored to user preferences for subscribed stocks from the Bombay Stock Exchange. Integrated 10+ commands utilizing Redis for efficient user data caching.
• Developed university navigation chatbot for course details and various college-related inquiries using tokenization techniques, Feedforward neural network, and LSTM models.

Samsung Research Institute, Bangalore

10/2020 – 03/2021   Remote
R&D Student

• Built low latency microservices to identify target devices for action, based on user utterances within 700ms on Samsung Bixby.
• Developed Flask APIs to generate sentence embeddings for user utterances by evaluating across 4 embedding models, and cached embedding vectors in Redis, improving latency by 25%.
• Implemented a time-decayed similarity algorithm to identify 3+ target devices based on user request frequency and latency, received an Excellence award for developing the services.

Mzaalo, Xfinite Global PLC

07/2020 – 08/2020   Remote
Growth Engineering Intern

• Built a Telegram chatbot for content sharing, overseeing the seamless integration of backend APIs for authentication and a reward system. Utilized Python and the Telegram API to achieve effective functionality and connectivity.

FindMind Analytics

04/2020 – 06/2020   Remote
Backend Development Intern

• Migrated the backend codebase to FastAPI and redesigned APIs for machine learning models, incorporating asynchronous operations with NoSQL database, leading to over 50% improvements in backend response times.
• Created web scrapers for documents in Nodejs and automated code deployment and scraper tasks through shell scripts.

Projects

Arxiv Sanity XL - Distributed Search Engine

Tech Stack: Java, Apache Zookeeper, FastAPI, PgVector, Postgres, Redis, Celery, Protobuf

• An event-driven, and highly available distributed document search engine leveraging leader election algorithm to elect cluster leaders for efficient workload distribution.
• Implemented the TF-IDF, and BM25 algorithm, generated document-chunk embeddings, and leveraged PgVector ANN retrieval to fetch required arxiv papers based on given keywords across a corpus of 30K+ research papers.

CodETL

Tech Stack: PySpark, Flask, Celery, Postgres, AWS S3, AWS EC2, Apache Airflow

• Spearheaded the development of a platform-agnostic ETL engine based on a standardized transformation map to spawn workers supporting 25+ transformations on the landing and staging layers.
• Created and integrated topologically sorted schedules to maximize parallel processing, orchestrating end-to-end ETL process on Airflow.
• Recognized with an Outstanding award by Deloitte for reducing development efforts by 40% over existing static SQL-based solutions.

DataLens

Tech Stack: Python, LangChain, LLM

• Developed an LLM-powered search engine and Q&A portal for structured and unstructured data using Retrieval Augmented Generation for postgres database, and documents.

Implementation of Genetic Algorithm for Path Traversal

Tech Stack: Javascript, P5.js

• Developed an interactive tool for obstacle creation and avoidance using path-finding operation in 2D environment employing Genetic Algorithm, inspired by Dan Shiffman’s book ”The Nature of Code.

Research

Publications

• EinsteinPy: A Community Python Package for General Relativity, Astrophysics Source Code Library, Bapat S., et. al.
• Cross-Geography Generalization of Machine Learning Methods for Classification of Flooded Regions in Aerial Images, Arxiv, Sushant L. et. al.
• Music Emotion Recognition Using CNN-LSTM Architecture With Hybrid Spectral - Temporal Audio Feature Vector, IEEE Access, Pratyush K., et al. (under review)
• Multiclass Classification and Verification of Online Signatures, Research project under Dr. Ramya S., Presented as part of Industrial Training at Manipal Institute of Technology. (A+ Grade)

Research Projects

Active Projects

• Autonomous Indoor Navigation
  
  • Comparative Study of Decision Transformer with Online Reinforcement Learning Methods for Autonomous Indoor Navigation.

Past Projects

• Music Emotion Recognition Using Hybrid Spectral - Temporal Feature Vector
  
  • Built a music emotion classification system based on Russell’s two-dimensional valence-arousal model using hybrid Convolutional Neural Network and Long Short Term Memory architectures.
  • Extracted a hybrid spectral-temporal feature vector involving audio fingerprints for training the model.
• Multiclass Classification and Verification of Online Signatures
  
  • Developed a robust signature verification tool that uses time-series data to detect forged signatures and classify signature owners using SVM.
  • Improved model efficiency by reducing the feature dimension using the Ramer-Douglas-Peucker sampling algorithm.

Achievements & Activities

Certifications

• AI certified by Deloitte AI Academy.
• Algorithmic Toolbox by University of California San Diego on Coursera.
• Algorithms on Graphs by University of California San Diego on Coursera.
• Machine Learning Foundations: A Case Study Approach by University of Washington on Coursera.
• Neural Networks and Deep Learning by Coursera.
• Combinatorics and Probability by University of California San Diego on Coursera.

Activities

• Trained 10+ new project recruits on core Python and PySpark concepts.
• Mentored students on core Machine Learning concepts at the Deloitte AI Academy.
• Provided software-based products and solutions to clients in over five countries as a freelancer.
• Published 100+ technical news articles on TechQuila, from Hardware Builds to Software News, read by over 700,000 users every month globally.
• Served as volunteer and organizer at TechTatva - National Technical Fest and Revels - National Cultural Fest, MIT Manipal.

Achievements

• Finished in top 3 among 70+ participants in Deloitte project-level hackathon.
• Received Outstanding Award by Deloitte for contributions towards the CodETL project.
• Received Excellence Award and $300 reward for developing intelligent ’Device Detection Algorithms’ during the Samsung PRISM internship program.
• Received 3 Applause Awards by Deloitte for contributions towards the Fortune 500 insurance services client and AIOPS.D project.