Apr 2022 – Jun 2023 · 1 yr 2 mos · Hyderabad, Telangana, India · Hybrid

• Building distributed microservices on top of gRPC and thrift.
• Build demo agent profile service from scratch for driving business growth.
• Took part in multiple initiatives for infrastructure optimisation to reduce cost.
• Worked on gRPC middleware modules to enhance application monitoring using DataDog tracing
• libraries. Took responsibility as admin for APM(Application Performance Monitoring) and log
• management domain and thus gaining knowledge around DataDog SaaS platform.

Nov 2020 – Mar 2022 · 1 yr 4 mos · Hyderabad, Telangana, India · Hybrid

• Developed toolchain for gRPC performance testing with locust, additionally also created pipeline for
• build and deployment of load test images to K8s.
• Gained understanding of codebase delivery to orchestration systems like Kubernetes under AWS,
• interacted with cloud services like IAM, S3, VPC, EC2, ECS, EKS, etc.
• Working with CircleCI for CI/CD pipelines, involving code build and upload process.

Jul 2019 – Nov 2020 · 1 yr 4 mos · Delhi Area, India

• Cloud Team at Samsung Research Institute - Delhi
• Worked on Samsung’s Core Smart TV services involving data processing pipelines with tech stack
• of Big Data tools like Apache Spark, Apache Kafka, Redis etc, with deployment on AWS EC2, S3, SQS, etc. Also gained knowledge of associated SDKs like boto3(python).
• Improved pipeline by using RQ(Redis Queue, a pythonic task queue with Redis as backend).
• Performance optimizations using parallelism and concurrency constructs like threading, multipro-cessing and asyncio. Removed IO based bottlenecks by implementing concept of pipelining in Redis.
• Created stream processsing pipelines using Faust(pure python based stream processing, similar to Kafka Streams).

Dec 2018 – Feb 2019 · 2 mos · Bangalore

• Sound Source Localization in Visual Scenes
• As humans we can localise the sound source using binaural mechanism but here the task involved sound localization without any information of DOA (Direction of Arrival) and only on basis of video feed.
• Worked on associative models which create association between certain objects and sound notes and analysed their strengths and weaknesses.
• Work done on state of the art research papers :
• 1 ) Created implementation of Learning to Localize Sound (CVPR 2018).
• 2 ) Developed implementation of Google DeepMind's paper "Objects That Sound" in Pytorch.

Jun 2018 – Aug 2018 · 2 mos · New Delhi Area, India

• Worked on Voice Classification to enhance device-user engagement and experience.
• A lot of IOT devices are now equipped with microphones, speakers, cameras etc. A smart gesture for a device can be to play some content based on current user's mood, age, gender etc.
• Under this project I worked to achieve such intelligence by predicting these parameters from audio note.
• Work done:
• 1) Developed a lightweight model capable of running on low-end hardware which could predict gender, emotion, based on an audio note.
• 2) Applied linear convolutional neural network over MFCC features and then improved performance by applying state of the art CNNs over specgrams of audio files.
• 3) Mainly worked on Tensorflow and Keras API for development.