2023 - 2025
Masters Degree
University of New Haven
GPA: 3.8/4.0
constcoder={name:'Bhaskara Vanacharla',skills:['React', 'Python', 'Dart', 'AWS', 'Javascript', 'MySql', 'Node.js', 'Docker'],hardWorker:true,quickLearner:true,problemSolver:true,hireable:function() {return(this.hardWorker&&this.problemSolver&&this.skills.length>=5);};};ABOUT ME
Who am I?
Hey there, I'm BHASKARA VANACHARLA! I'm a developer who finds genuine joy in writing clean, maintainable code and solving complex problems. Whether it's debugging at midnight or exploring emerging technologies, I approach every challenge with curiosity and determination. My philosophy? Learn by doing, iterate constantly, and never stop growing.
I take pride in writing code that's not just functional, but elegant and well-documented. My strength lies in transforming complex systems into intuitive solutions that users love. I believe in the power of good architecture, comprehensive documentation, and code that tells a story. Currently seeking opportunities where I can contribute to meaningful projects and collaborate with teams that value quality, innovation, and a touch of creativity.
EXPERIENCE
General Neuro
constjob={myRole:Software Engineer,duration:(Jan 2025 - Present),tools: ['Flutter', 'Dart', 'Firebase', 'OpenAI API', 'Gemini API', 'Sentry', 'OOP', 'Spaced Repetition],Description: Developed a production-scale Flutter companion app for the NeuroLingo brain stimulation headset supporting 200+ active users. Pioneered AI-powered learning workflows with cloze/MCQ generation and spaced-repetition flashcards using OpenAI and Gemini APIs, supporting multilingual learning across 12 languages (~300k records). Reduced app load time by 95% (from 30-60s to 2-3s) by migrating JSON bundles from Firestore to Firebase Storage and integrated Sentry for error tracking, achieving a 20% reduction in post-release failures.};
University of New Haven
constjob={myRole:Graduate Assistant and Teaching Assistant,duration:(Sep 2024 - Jan 2025),tools: ['Python', 'PyTorch', 'Machine Learning', 'Neural Networks', 'NLP', 'MPS Acceleration', 'Predictive Modeling],Description: Built an end-to-end admissions prediction pipeline using PyTorch neural network with MPS acceleration, cutting training time by 50% while maintaining a lightweight model (~6 MB, 13K parameters). Integrated the ML model into admissions workflows, isolating 1,055 high-likelihood enrollees (~8.4% of 12,569 applications) with 81.9% recall, reducing manual review scope by 90%. Mentored 50 students in Natural Language Processing, delivering tutoring sessions and hands-on learning in ML algorithms.};
Epam
constjob={myRole:Software Engineer,duration:(Jul 2022 - Oct 2023),tools: ['React.js', 'Redux', 'Node.js', 'Express', 'RESTful APIs', 'Jest', 'Mocha', 'Agile', 'Jira],Description: Engineered scalable React.js applications with Redux state management, delivering 40+ reusable UI components and leading UI redesigns across 6 internal web applications serving 10,000+ daily users. Designed and deployed 15+ RESTful backend services using Node.js and Express, improving cross-platform data exchange by 25% and achieving 99.9% uptime. Decreased application load times by 40% through React.js codebase refactoring, hooks optimization, and memoization. Increased code coverage to 90% using Jest and Mocha testing frameworks.};
Cognizant Technology Solutions
constjob={myRole:Software Engineer,duration:(Jun 2021 - Jun 2022),tools: ['React.js', 'TypeScript', 'AWS AppSync', 'DynamoDB', 'Cognito', 'Redux', 'Node.js', 'Express', 'Docker],Description: Built a scalable online bookstore using React.js and TypeScript with an AWS backend (AppSync, DynamoDB, Cognito), providing real-time inventory updates, advanced search and filtering, persistent wishlist and cart, user authentication, and multilingual support — reducing average checkout time by 30%. Managed React components with Redux to boost developer productivity by 50–60% and increase code reusability. Implemented server-side logic with Node.js and Express, improving response times by 15%. Containerized services with Docker to standardize environments and reduce setup time by 40%.};Skills
PROJECTS
LLM Vulnerability Scanner
constproject={name:'LLM Vulnerability Scanner',tools: ['Python', 'FastAPI', 'Streamlit', 'Docker', 'Ollama', 'Llama Guard', 'Security Testing],Description: Developed a comprehensive security testing tool for Large Language Models using FastAPI and Streamlit. Implemented automated vulnerability scanning leveraging Llama Guard to detect potential security risks, prompt injections, and unsafe outputs. Containerized the application with Docker for seamless deployment and scalability across different environments.};YOLOv5s Parking Space Classifier
constproject={name:'YOLOv5s Parking Space Classifier',tools: ['Python', 'PyTorch', 'YOLOv5', 'Computer Vision', 'Flask', 'Google Colab', 'Git', 'Pattern Recognition],Description: Built a real-time parking space detection system using YOLOv5s object detection model trained on custom datasets. Implemented computer vision techniques for pattern recognition to classify occupied vs. available parking spots with 95% accuracy. Deployed the model using Gradio, enabling real-time inference for parking management systems.};Tic-Tac-Toe AI with Minimax and MCTS
constproject={name:'Tic-Tac-Toe AI with Minimax and MCTS',tools: ['Python', 'Minimax Algorithm', 'Monte Carlo Tree Search', 'Game Theory', 'AI', 'Algorithm Optimization],Description: Implemented and compared two AI algorithms for Tic-Tac-Toe: Minimax with alpha-beta pruning and Monte Carlo Tree Search (MCTS). Analyzed algorithm performance characteristics, with Minimax providing deterministic optimal play through exhaustive game tree search, while MCTS demonstrated probabilistic decision-making through random simulations. Evaluated computational efficiency, move selection speed, and decision-making patterns across different game states to understand the trade-offs between both approaches.};Adaptive Multimedia Streaming in Wireless Networks
constproject={name:'Adaptive Multimedia Streaming in Wireless Networks',tools: ['NS-3', 'C++', 'MPEG-DASH', 'WLAN', 'HTTP', 'Network Simulation', 'Quality of Service', 'UDP', 'RTP', 'LTE],Description: Simulated adaptive bitrate streaming protocols using NS-3 network simulator to optimize multimedia delivery across multiple wireless network topologies. Implemented MPEG-DASH standard with buffer-based rate adaptation algorithms to dynamically adjust bitrate based on network conditions and buffer occupancy, reducing rebuffering events by 40% during steady-state operation. Evaluated multiple streaming protocols (UDP, RTP, DASH) across single-hop and mesh network configurations, achieving 30% improvement in average video quality and effective throughput in high-congestion scenarios through accurate capacity estimation during start-up phase and buffer management in steady state.};Retinal Disease Classification on OCT Images
constproject={name:'Retinal Disease Classification on OCT Images',tools: ['Python', 'TensorFlow', 'Keras', 'Xception', 'ResNet50', 'Transfer Learning', 'Medical Imaging', 'Deep Learning', 'Data Augmentation],Description: Developed a deep learning model for automated retinal disease classification using Optical Coherence Tomography (OCT) images to detect AMD, DME, and CNV pathologies. Leveraged transfer learning with Xception and ResNet50 pre-trained architectures, achieving 96.8% accuracy on 84,000+ annotated SD-OCT images - outperforming traditional handcrafted feature extraction methods (HOG/LBP) by 46%. Applied data augmentation techniques and optimized train-validation splits to prevent overfitting, demonstrating superior performance compared to existing CNN-from-scratch approaches while enabling early disease detection for improved patient healthcare outcomes.};ACHIEVEMENTS
TechCrunch Disrupt 2024 — Battlefield Attendee (Top 200 selected teams)
``
Project Demo at Google NYC (AI for Startups)
``
Founding Team Member (Software Engineering) – General Neuro
``
Education
2018 - 2022
Bachelors Degree
Anurag Group of Institutions
CGPA: 8.5/10
