Experience

Graduate Student Researcher | SOPAC Lab

Location: Scripps Institute of Oceanography, UC San Diego, CA, USA

Duration: Jan 2023 - Present | 2 mos

ML Modeling: Developing a ML pipeline in collaboration with NASA JPL, to classify spatio-temporal anomalies. Executed exploratory data analysis for feature extraction and conducted experiments with random forests and GCNs.
MLOps and Distributed Infrastructure: Implemented distributed training on Kubernetes, enhancing scalability for large datasets. Experimented with various ML architectures and performed hyperparameter tuning, improving model accuracy.
Real-Time Data Processing Software: Developed a multi-threaded Qt-based software, enabling real-time processing of sensor data from over 100 servers, incorporating APIs for data correction and GPS-based user interaction.
Scalable Server and GUI: Engineered a scalable server capable of handling simultaneous client connections with secure user authentication. Designed an intuitive GUI for monitoring client statuses and managing server connections.
Data Processing: Developed Python scripts to validate geodetic data spanning over three decades, using recorded earthquake events and sensor fault/change events, resulting in accurate data classification into outliers, transients, and artifacts for further analysis.

Door-key Problem

Location: Hyderabad, Telangana, India

Duration: July 2021 - Aug 2022 | 1 yr 2 mos

Edge Computing: Spearheaded Nayanamv2 development, a GPU-based vehicle tracking system on edge devices, utilizing deep learning models to achieve accuracy between 85-97% based on streaming needs. Integrated socket server for client data broadcasting and executed batch inference deployment of a multi-stream processing model on Nvidia Jetson.
Vehicle Counting Software: Conceptualized and developed Nayanamv1, a real-time image processing solution. Designed for low-resource environments, this system accurately counts vehicles at stop lines using live stream.
Traffic Management Software Development: Led TIMv2 and TIMv3 projects enhancing autonomous functionality, traffic signal optimization and detector integration with improvements in server communication, traffic flow algorithms and system reliability. Worked in docker environments for testing software in traffic simulation.
Server and Application Development: Developed server-side software with Redis, gRPC and FastAPI for efficient database integration and client communication. Built a Flutter-based Enforcement App for highway incident monitoring.

Door-key Problem

Location: Hyderabad, Telangana, India

Duration: Apr 2020 - Jul 2020 | 4 mos

API Development: Engineered a solution to optimize traffic flow for emergency vehicles. Utilized geospatial analysis to identify traffic signals along the shortest route, enabling uninterrupted ambulance and police passage through junctions.
Backend Development: Authored APIs focusing on navigation and alert systems for buses. Implemented algorithms for shortest path calculations and developed real-time geo-fencing, bunching alerts to enhance public transport efficiency.
Mobile App Development: Deployed Open Trip Planner on an AWS instance for public transport and walking route planning. Contributed to the Margadarsi app using Flutter, focusing on features like journey planning and bus timetables.

Door-key Problem

Location: Kochi, Kerala, India

Duration: Dec 2019 - Jan 2020 | 2 mos

Developed a Blockchain architecture using Hyperledger-Sawtooth to secure data flow in a ‘manufacturer-maintenance’ system.

Door-key Problem

Location: Hyderabad, Telangana, India

Duration: May 2019 - Jul 2019 | 3 mos

Developed a voice controlled toy car using logistic regression algorithm.
The code is written from scratch using basic principles and the voice instructions are processed on raspberry pi and sent over a Bluetooth channel to Arduino placed on the toy car.

Door-key Problem