Jun 2023 – Mar 2025 · 1 yr 9 mos · Gurugram, Haryana, India · On-site

• Part of the Zomato Support team; supervised, built, and maintained various features related to Chat Bot and Internal Support Tool for smooth issue resolution for customers and support executives catering to over 18 million+ active users.
• Implemented end-to-end image resolution for support for spilled food from Chat Bot for 10 different cuisine types using GPT 4o Vision, Kafka, gRPC, Worker, and Scheduler, prompting across 3 different services resulting in a monthly cost saving of 1 million rupees & a 20% increase in customer satisfaction.
• Led Zomato Conversational AI Bot, implemented new flows, bug fixes, improved stability, implemented fallback model handling and circuit breaker, added end-to-end monitoring and alerting across 3 different services resulting in a monthly cost saving of 10 million rupees and 30% increase in customer satisfaction. Used different models ranging from GPT 3.5, GPT 4o mini, GPT 4o, GPT 4o fine-tuned, Llama, and Claude 3.5 across different deployment providers: OpenAI, Azure, Together.ai, and AWS.
• Implemented real-time recoupment flow from the Restaurant in case a refund was made via chatbot using SQS, Kafka, Scheduler, and RDS, resulting in a monthly cost saving of 1.8 million rupees.
• Added Order Summary PDF entity sum validation and fixed 10 critical order total dispute bugs to ensure customers receive correct data. Added Monitoring and Alert using Grafana, Prometheus, PagerDuty, and Jumbo Logging for any discrepancy.
• Implemented Customer Escalation Flow in support bot to connect agent to customers, reducing Resolution time by 70% and increasing customer satisfaction by 150% in affected 10k+ cases daily.
• Identified and fixed major security issues in the Internal Support Tool regarding User PII information. Added rate limiting, new permission groups, and User PII Unmasking Limits on User Level, restricting access by 80% for all users.
• Implemented Callback functionality in Chat Bot, increasing customer satisfaction by 20%.

Jan 2023 – Jun 2023 · 5 mos · Bengaluru, Karnataka, India

• Collaborated with Tekton CI/CD team, a Kubernetes-based CI/CD Pipeline using Go, and OpenShift Pipeline service. Shipped Tekton Task for Skopeo-Copy, enabling secure Docker image backups.
• Ensured End to End Testing, GitHub Action Release & Customer Support for Skopeo-Copy, Git, Buildah.

Jun 2022 – Jul 2022 · 1 mo · Bengaluru, Karnataka, India

• Changed Data Model of Guidance Panel of the COBN portal by combining all data models into a single data model and parsed data using Sybase SQL, Reladomo, Java.
• Created REST API to fetch guidance from the new data model and used Redux-Saga to update the states on frontend based on React JS.
• Added tabs, collapse sections in Frontend to make guidance panel more structured and easier for users to navigate through it.

Mar 2022 – Sep 2022 · 6 mos · Remote

• Successfully Mentored 4 projects under Radis, performed code reviews, resolved issues, and guided mentees throughout the program.
• Became 3rd highest contributor with 130+ commits. Published a new version on PyPI and drafted a new release on GitHub.

Sep 2021 – Feb 2025 · 3 yrs 5 mos · Remote

May 2021 – Aug 2021 · 3 mos · Remote

• Developed an optimum Lineshape solution for Radis Radiation increasing performance by 40x times. Wrote 4400+ lines of code in Python becoming 2nd highest overall contributor.
• Implemented Profiler to store/print parameters, methods along calculation time (90% code-coverage).
• Added wstep=’auto’ method by refactoring the codebase to precompute optimum value of 'wstep' parameter maintaining accuracy and performance.
• Identified bottlenecks in the current implementation, derived/verified time complexity of all methods by running 50+ benchmarks based on Tableau and Xexplorer(An interactive plotting library based on Excel + Pandas + Bokeh + Holoviz).
• Replaced existing config file from 'ini' format to 'JSON' format with backward compatability.

Mar 2022 – Apr 2022 · 1 mo · Boston, Massachusetts, United States

• Worked on budget prediction model using 7 different parameters of a movie script based on its story adaptation.
• Preprocessed 560+ data, trained model using different loss functions using scipy optimize library (rosa) and custom Root Mean Square Error function, and compared performance with baseline method.
• Annotated over 1500+ sentences for 10 Cinematic Syntax Assimilation Parameters for different indicators analyzing Sentimentality with respect to CSA parameter.

Feb 2022 – Jun 2022 · 4 mos · Remote

• Mentored 2 open-source projects - Twitter-video-emotion-and-sentiment-analysis and Python Assistant and became a core maintainer for the repositories.
• Performed code-reviews, resolved issues, guided mentees throughout the program.
• Updated Readme and contirbuting guidelines to make project more systematic and easier for mentees to contribute to the project.

Mar 2021 – May 2021 · 2 mos · Remote

• Contributed to the 'EmotionRecognitionFromFacialExpressions' repository, added code to automatically download the dataset from Kaggle using API Token.
• Added code for generating Confusion Matrix and created environment file.
• Contributed to the 'FaceX' repository, improving documentation and Face detention algorithm.

Jan 2021 – Mar 2021 · 2 mos · Singapore

• Made diverse Visualizations and Data Analysis primarily based on 1 Million+ podcasts data using Tableau by doing data extraction using Python.
• Gave insights on optimum frequency and consistency of episodes of podcasts.
• Built a prototype model for AWS Transcribe using Django and examined AWS Codeguru’s compatibility with the existing codebase.
• Improved features of App2 Website. Increased accuracy/performance of NLP model by 15% by using new Web Scraping technique.

Sep 2020 – May 2022 · 1 yr 8 mos

Aug 2019 – Aug 2020 · 1 yr

• Conducted Workshop for 1st Year Students on AI Obstacle Avoidance Bot which saw a participation of 300+ students.
• Built AI Quadruped Bot Project for Technical Fest Nimbus.
• Collaborated on Leap Motion Bot project in a team of 2 where we used Arduino and Leap Motion to control a 4 Mecanum wheels Bot using hand gestures.
• Gave a virtual Lightning Talk at an Open Source Event conducted by GLUH-NITH about Open Source Contribution and GSoC which saw a participation of 200+ students.

May 2020 – Aug 2020 · 3 mos · Bengaluru, Karnataka, India

• I was responsible for creating 3 OCR models using PyTesseract, Google Vision and AWS Textract
• for extracting various blood tests and their values from CBC reports, Urine tests, etc.
• Designed a web scraper for searching websites and extracting Causes, Symptoms, Treatment, and Diagnoses of any disease and also finding medicines for the same.
• Creating a recommendation system in which we will get 3 inputs from the user. Their Blood report and by that OCR we built we will find abnormal values of blood tests. We had a chatbot in which the user will tell us the symptoms he or she is having Polyuria, Fatigue, etc. Manual Inputs of Blood Pressure, Height, Weight, HBA1C, etc.
• Based on the severity of the problems the model would give recommendations like stop consuming Alcohol, Do yoga to cope with stress, seek a doctor, etc.