Aug 2024 – Dec 2024 · 4 mos · Bengaluru, Karnataka, India · On-site

• Customer Support Chat Service
• Designed and developed a backend chat service in Node.js using an HTTP API and upgrading the connection WebSocket for real-time communication, with session management and conversation context stored in Firestore.
• Led data preprocessing, trained and integrated Google Dialogflow for automated responses and backend API calls, escalating unresolved queries to live CX agents via Freshchat REST APIs.
• Preserved conversation context using a custom data structures for efficient retrieval and escalation.
• Leveraged Freshchat webhooks using Firebase Functions to handle events like message updates and agent assignment in real-time.
• Deployed the solution on Google Cloud Services, ensuring scalability and seamless integration of third-party APIs.
• Spend Analyzer
• Designed and implemented a spend analysis service to extract transactional data from user messages, storing and managing data in Firestore for efficient retrieval and processing.
• Built a Firebase Functions GET API to provide spend details, leveraging template-based parsing to analyze messages and handle diverse transaction scenarios.
• Developed mechanisms to eliminate false positives, addressing edge cases such as self-transfers, refunds, ATM withdrawals, self-cheques, and detecting multiple messages for the same transaction.
• Delivered enhanced user experience by generating personalized animated videos showcasing spending insights, driving user engagement.

Feb 2024 – Jun 2024 · 4 mos · Bengaluru, Karnataka, India · On-site

• Part of the Inferred Planogram and AEG teams under Just-Walk-Out, AWS.
• Architected and implemented the Metrics Emission framework for Associate tools with AWS services for real-time logging, transformation, and storage, and developed a CI/CD pipeline with AWS CDK in TypeScript for infrastructure management and automation.
• Used AWS Glue for schema management, S3 for Parquet storage, Athena for querying, and QuickSight for visualization.Integrated CloudWatch for monitoring, logging, and alerting, ensuring operational reliability.
• Contributed to Error Triaging tool with a React-based frontend, seamlessly integrating backend services using Java, managed infrastructure using AWS CDK in TypeScript.
• Engineered a high-performance DynamoDB schema, integrated with AWS Lambda for real-time processing, optimized queries with partition and sort keys, and exposed functionality via a GET API.
• Leveraged Apache Parquet for optimized data storage in S3 bucket, adhered to best practices such as DAO patterns and design principles in Java to ensure maintainability and scalability.
• Resolved high-impact SAS risks through advanced debugging, dependency graph analysis, and deadlock resolution, enhancing system integrity and security
• Added comprehensive unit tests using JUnit 5, ensuring high code quality and stability.

Sep 2022 – Dec 2022 · 3 mos · Lucknow, Uttar Pradesh, India · Remote

• Designed and developed a methodology for detecting and counting sugarcane tillers using computer vision in Python and TensorFlow. Conducted data collection and annotation with open source tool, Labelme.
• Fine-tuned a YOLOv7 object detection model for accurate tiller detection achieving 92.64% mAP.
• Developed and containerized a FastAPI server that processes input images with the YOLOv7 model, returns bounding boxes and tiller counts, and ensures consistent deployment using Docker.

Jun 2022 – Jul 2022 · 1 mo · Gwalior, Madhya Pradesh, India · Remote

• Compared PSO, GA, WOA metaheuristic algorithms for soil fertility dataset's cost function optimization.
• Used them for ANN hyperparameter tuning.

Mar 2021 – Jan 2022 · 10 mos · Raipur, Chhattisgarh, India

• Specialization in statistics and data analysis.
• Addressed and resolved queries from high school and university students.