Results-focused software engineering professional with extensive experience in development, machine learning and
computer vision. Analytical and detail-oriented.
Overview
4
4
years of professional experience
Work History
Software Development Engineer
Flipkart
Bengaluru
04.2022 - 03.2024
Worked as a part of pricing where I primarily owned matching service while working on other component like crawling, API push manager and ingestion pipeline.
Spearheaded the execution of a product match recommendation service improvement plan, enhancing overall system accuracy and performance.
Conducted exploratory analysis to identify and rectify root causes of accuracy issues, significantly boosting process efficiency.
Developed a transformer-based similarity search using FAISS indexing for accelerated data scanning.
Proposed a high-level design (HLD) for workflow automation within the matching pipeline.
Enhanced the Human-in-the-Loop (HITL) process, improving the operational experience for users.
Led the redesign of the crawling system, integrating Celery and RabbitMQ for enhanced data retrieval capabilities.
Designed the API push manager service to facilitate client-side publication of reprice results.
Optimized data ingestion workflows by implementing dynamic feed file ingestion alongside master feed ingestion.
Software Development Engineer
Samsung R&D Institute Delhi
01.2020 - 04.2022
Developed an on-device fitness app featuring real-time multi-person body pose tracking using a quantized OpenPose network, optimized with KL-Divergence for enhanced precision.
Implemented a C++ API for parsing human keypoints using non-maximum suppression and a k-partite graph for body part associations.
Utilized a disjoint set data structure to accurately determine full body poses of multiple individuals.
Refined body pose detection by incorporating physical and geometric constraints, tailored for a multi-person fitness application.
Engineered a C++ and Python API for on-device multiple object detection leveraging the neural processing unit (NPU).
Enhanced YoloV4, YoloV5, and EfficientDet models through pruning, graph optimization, and quantization, reducing latency by 70% and surpassing FPS targets by 30% with maintained accuracy.
Education
B.tech - EEE
Indian Institute of Technology Patna
07.2020
Skills
C/C
Python
Javascript
Typescript
Java
ReactJS
NextJS
GraphQL
MySql
Azure
MongoDB
Jquery
AngularJS
HTML5/CSS3
Accomplishments
Candidate master on Codeforces. Ranked 184th in India out of 10k users.
Ranked 223rd in Atcoder in India out of 10k users.
Secured 96.9 percentile in JEE Advanced Exam out of 0.2 million Candidates.
Cleared SWC test in first attempt. SWC stands for Samsung Software Competency Test conducted by Samsung and is a 4 hours coding round.
Awarded instant appreciation award in Flipkart for outstanding performance in match improvement
Hobbies and Interests
Competitive Programming
System Design
Data Structures and Algorithms
Major Project
Developed a fully auditable movie summarizer service, integrating TTS APIs, CLIP model for image matching, and an audio speech recognition pipeline to enhance semantic understanding of movies. Employed the MoviePy library for efficient movie summarization.