Varun Tarikere Shankarappa

Teaching Assistant, Manning College of Information and Computer Science, UMass, Amherst

June 2024 – Nov 2024

• Evaluated and graded assignments, projects, and exams, ensuring accurate assessment of students' understanding of software engineering principles(COMPSCI-520).
• Provided constructive feedback to students, helping them improve their coding practices, design techniques, and problem-solving skills.

Graduate Student Researcher, Human Centered Robotics Lab , UMass, Amherst

June 2024 – Nov 2024

• Optimized a model for frame-level fault detection in collaborative autonomous systems, processing 30 images per timestep from 5 agents in the DeepAccident dataset.
• Conducted an in-depth analysis of dataset structure and identified key challenges impacting detection outcomes.
• Designed and implemented a denoising-based fault detection model with Vector Quantized GANs and Graph Attention Networks as pre-training, exploring enhanced learning for fault detection in multi-agent autonomous systems.
• Optimized a non-linear MPC-WBIC controller through semi-black box fine-tuning, tailoring it for improved performance on the Go1 robot platform.

Dining Hall Supervisor (Part-Time), Auxiliary Enterprises, UMass, Amherst

Sep 2023 – Dec 2024

• Handled 2000+ students per meal session
• Led and motivated team members to maintain high performance and productivity, ensuring smooth operations across all work areas.
• Strategically managed employee break schedules to balance workload and optimize efficiency.
• Demonstrated strong attention to detail by ensuring allergens were accurately identified and communicated, fostering a safe and compliant work environment.

Software Development Engineer - I, Eagleview, Bengaluru

Aug 2020 – Dec 2022

• Assisted in designing and led implementation of an event-based notification service handling push, email, and iOS notifications via AWS SQS and SNS, supporting 1,000+ concurrent requests.
• Gained proficiency in C# and delivered critical release APIs within one week.
• Delivered 6+ Golang and Python microservices, collaborating with cross-continental agile teams.
• Contributed to migrating a legacy platform to a serverless orchestrator using AWS Step Functions and Lambda, reducing service time from 3 hours to 45 minutes.
• Achieved 60% latency reduction for large file downloads by implementing gRPC stream APIs.
• Enhanced service performance and resolved critical bugs through regular load and performance testing.
• Translated business-critical MATLAB code handling geospatial data to Python, increasing scalability and reducing latency by over 70%.
• Participated in the full SDLC of Assess, an online imagery inspection tool with a $1M+ projected revenue.
• Proposed OpenTelemetry integration for serverless tracing, improving logging and debugging capabilities.
• Participated in code reviews to ensure clean, tested, readable and maintainable codebase to maintain 90% code coverage.

Software Development Engineer Intern, Eagleview, Bengaluru

Jan 2020 – June 2020

• Delivered 15+ APIs for a critical process and improved API documentation, reducing cross-team dependency and cutting development time by 15%..
• Presented a study on ElasticSearch-Logstash-Kibana (ELK) stack for observability, demonstrating its potential to trace points of failure within 30 minutes and deliver key analytical insights.
• Proto-typed a dashboard to automate version tracking, cutting manual checks and documentation time by 50%.

Research Intern, Center for Cloud and Big Data, PES University, Bengaluru

May 2018 – Aug 2018

• Researched on federated clouds using OpenStack.
• Conducted a feasibility study of federated container service using Mesos in Openstack.

Distributed Rate Limiter

• Designed a gRPC-based rate-limiting system using Redis for storage.
• Implemented a FastAPI-based dummy service to validate request throttling.
• Developed a Kubernetes operator to dynamically configure rate limits via CRDs.
• Deployed services using Docker and Kubernetes with automated policy enforcement.

Human Following Robot

• Developed a sophisticated human-following capability for a holonomic robot equipped with an RGB-D camera.
• Leveraged advanced machine learning models such as YOLO-v7 for object detection, SuperPoint for feature extraction, and a PID controller for precise movement.
• Integrated computer vision and control systems for real-time human tracking in dynamic environments.

Quote Classification and Interpretation

• Fine-tuned Transformer models like BERT, GPT-2, and T5 for multi-label classification of quotes.
• Explored prompting techniques on T5 and Gemma models to enhance quote interpretation.
• Improved model accuracy and efficiency through extensive analysis of results.

Stock Exchange Application

• Prototyped a stock trading application in Python to experiment with distributed systems concepts.
• Implemented multithreading and leader elections for a realistic trading simulation.
• Tested algorithms for fault tolerance and system optimization.

Optimizing Machine Learning Models for Predictive Analysis

• Developed custom machine learning models, including Neural Networks, Random Forests, and K-Nearest Neighbors.
• Applied hyper-parameter tuning and optimization to maximize model performance.
• Generated F1 Score graphs for data-driven decision making.

Impact of Hofstede’s Cultural Dimensions on Intelligent Ethical Agent

• Explored moral dilemmas in autonomous vehicles, training AI models for ethically acceptable decision-making using data from the Moral Machine experiment.
• Evaluated model performance by analyzing accuracies across multiple training techniques and methodologies.
• Investigated cultural influence on AI decisions by incorporating Hofstede's six dimensions of national cultures during data preprocessing.