Ullas Basavapatna Chandrashekar.

Backend Engineer · M.S. CS @ GWU

Education

The George Washington University

Master of Science in Computer Science

August 2024 - May 2026
Washington, D.C.

VTU Bangalore

Bachelor of Engineering in Electronics and Communication

August 2018 - July 2022
Bangalore, India

Experience

Systems Engineer

Infosys Ltd.

• Designed and deployed modular microservices tailored for Belgium and UEFA-specific features, enabling dynamic rollout across 6 European countries; managing 12+ region-based feature flags and accelerated time-to-market by 2 weeks per release
• Developed internal developer tools and backend data validation scripts to streamline integration testing across ~30K simulated user sessions per environment; decreased pre-release QA overhead by 60+ engineering hours per cycle and improved sprint reliability
• Refactored legacy configuration modules and environment properties for over 22 microservices, cutting manual intervention in release workflows; led to near-zero configuration mismatches and consistent rollout across 6 test beds, supporting ~1.2M weekly active users

August 2022 - June 2024
Bangalore, India

Software Intern

Transo

• Built a real-time analytics dashboard for a logistics firm, integrating APIs to track 5,000+ daily shipments. Enabled deep insights into delivery status, driver scores, route efficiency, and inventory trends, cutting down transit delays by 18% through dynamic visualisations
• Led R&D on carbon footprint reduction, developing a predictive ML model (accuracy: 0.87), analysing vehicle age and engine efficiency. Informed fleet upgrades cut CO₂ emissions by 200+ metric tons/year, equivalent to 50 delivery trucks

August 2021 - October 2021
Bangalore, India

Skills

Programming & Scripting

Java
Python
SQL
C++
Ruby

Frameworks & Libraries

SpringBoot
React
NextJS
NodeJS
Rails

Databases

PostgreSQL
MySQL
MongoDB

Performance Testing

JMeter
LoadRunner
Performance Engineering
k6
Grafana
Kibana

Tools & Others

Git
Github
Linux/Unix
Jira
Agile
AWS

Projects

Semantic Drift Detection — NLP Pipeline for Emerging Slang & Language Change Analysis

Built an NLP system to detect semantic drift and emerging internet slang across Reddit communities over time. Designed a longitudinal language analysis pipeline using contextual embeddings, breakout detection, and clustering to identify when words appeared, how their meanings shifted, and which new senses emerged. Implemented large-scale subreddit data processing, temporal frequency tracking, contextual similarity analysis, and masked language modeling evaluation to estimate when evolving vocabulary could impact model understanding and signal the need for selective retraining.

Sentence Transformers
RoBERTa
HDBSCAN
Pandas
NumPy

WeCureIT — Full-Stack Medical Clinic Management & Intelligent Scheduling Platform

Designed and implemented an end-to-end medical clinic management system supporting patients, doctors, and administrators. Built a scalable full-stack platform with dynamic appointment scheduling, multi-facility doctor assignment, and rule-based break optimization. Implemented secure authentication, role-based access control, encrypted payment handling, and real-time availability computation. Integrated data-driven scheduling logic to prevent conflicts, enforce clinical constraints, and improve operational efficiency across healthcare facilities.

Next.js
Spring Boot
PostgreSQL
Firebase
Docker
Neo4j
Ollama

Driftline — Behavioral Drift Detection Platform

Built a behavioral drift detection platform over distributed traces, detecting dependency and p95 latency regressions (>2×) within 1–2 minutes using execution-graph diffing and baseline snapshots, processing 1,000+ spans/min with less than 100 ms ingestion latency and reducing simulated incident MTTR by ~50%.

Python
OpenTelemetry
Distributed Systems
Microservices
Docker

Obesity Risk Analytics — Spatiotemporal Modeling & Policy Decision Support System

Developed an end-to-end data analytics and machine learning framework to assess, forecast, and prioritize county-level obesity risk across the United States using multi-year public health and socioeconomic data (2010–2023). Conducted exploratory spatial analysis and unsupervised clustering to identify latent regional risk profiles, followed by supervised modeling and temporal trend analysis to capture evolving obesity dynamics. Designed early-warning indicators and long-horizon forecasting models to detect accelerating risk, and integrated results into a policy decision support dashboard for interpretable, data-driven intervention planning.

Python
Pandas NumPy
Scikit-learn
XGBoost
GeoPandas
SHAP
Plotly

AI-Powered Diagnostic Support Tool

Developed a medical diagnostic support system that utilizes Agentic RAG to suggest diseases and tests based on patient symptoms. By indexing over 230,000 PubMed abstracts into a ChromaDB vector store, the system ensures all AI-generated suggestions are backed by peer-reviewed evidence, reducing the risk of misinformation common in standard LLMs, and providing clinicians with transparent, data-driven insights for informed decision-making.

Node.js
React

Animal Intrusion Detection & Rescue System — Computer Vision & Embedded

Designed and implemented a real-time animal intrusion detection and rescue system using deep learning and embedded hardware. Integrated YOLO-based object detection with convolutional neural networks to identify wild animals from long-range surveillance feeds, achieving reliable detection up to 250 meters with 84% accuracy. Trained and fine-tuned the model on a custom dataset of over 5,000 annotated images to improve detection robustness across varied environments. Deployed the system on Raspberry Pi, optimizing video processing pipelines to reduce end-to-end latency to under 2 seconds for real-time monitoring and rapid response.

Python
YOLO
CNN
OpenCV
Raspberry Pi
Embedded Systems

Catg Programming Language — Compiler Design & Language Engineering

Designed and implemented a custom programming language supporting first-class functions and functional-style list operations within an imperative paradigm. Built a complete compiler pipeline from scratch, including lexical analysis, parsing, semantic handling, and runtime execution. Emphasized expressive functional constructs to reduce boilerplate, achieving approximately 30% reduction in code verbosity for functional-style programs. Validated language correctness and robustness by executing and testing over 100 custom programs across diverse control-flow and data-manipulation scenarios.