YashwantBezawada
Senior Tinkerer
I tinker with all kinds of nerdy things - AI, robotics, web apps, automation, you name it. If it's interesting and solves a problem, I'm probably building something with it. Technology is just a tool, curiosity is the real driver!
About
I build stuff with AI and it's pretty awesome
Hey there! I'm an AI Engineer who genuinely gets excited about teaching machines to do cool things. Currently working at FedEx, where I help build AI systems that make life easier for millions of customers.
What I love most? Taking those mind-blowing AI research papers and turning them into actual products people can use. Whether it's fine-tuning language models, building ML pipelines, or just making AI do something unexpectedly clever - that's where you'll find me geeking out.
AI Expertise
ML, Deep Learning, LLMs
Full Stack
Python, TypeScript, React
Production Ready
MLOps, Cloud Architecture
Team Leader
Agile, Mentorship
Education
Master of Science in Computer Science
George Mason University
Fairfax, VA • 2018 - 2020
Focus on Machine Learning and Artificial Intelligence
Bachelor of Technology in Computer Science
KL University
Vijayawada, India • 2014 - 2018
Research in Unsupervised Learning and Computer Vision
"I believe AI should make life simpler, not more complicated. Let's build something amazing together!"
Professional Experience
FedEx Corporate Services, Inc. | Collierville, TN
- Developed a Gen AI–powered customer-facing chatbot for shipment tracking, scheduling, and account queries that significantly increased user engagement and accelerated response times.
- Led a team to develop a multimodal retrieval-augmented generation (RAG) chatbot integrated with the FedEx knowledge base, streamlining access to packaging standards and enabling employees to quickly locate relevant documentation, cutting helpdesk resolution times.
- Collaborated with product and infrastructure teams to define a generative AI roadmap, contributing to operational efficiency improvements and driving innovation in associate tools and automation.
- Architected end-to-end MLOps pipelines to streamline deployments, enhance reliability, and enable server inference for continuously re-trained models.
- Migrated mission-critical applications to Azure Kubernetes Service (AKS), reducing deployment times by 65% and strengthening security measures, saving $800K annually.
Technologies:
FedEx Corporate Services, Inc. | Collierville, TN
- Developed and maintained global enterprise-level pricing solutions that impact internal pricing users and external customers across the globe to support FedEx's business strategy and goals with a customer-first ideology
- Spearheaded multiple application modernization efforts including Legacy to CloudOps Migration, WebLogic to Spring Boot Conversion, Spring Boot Admin, API Gateway and Service Discovery adoption
- Developed Enterprise Discount Database (EDD) Archive-Purge process dramatically reducing the Database growth rate in GBs and cutting down the overall DB size by 11 TB
- Modernized legacy applications by migrating them to cloud-native microservices, cutting maintenance costs by 20% and reducing system downtime by 30%
Technologies:
Navy Federal Credit Union | Vienna, VA
- Responsible for refactoring existing disputes system using Payment and Dispute Management Suite from Visa and Verifi
- Implementing software for customer-facing products and internal corporate tools
- Part of the NFCU Digital Labs team using Agile methodologies, including Scrum, pair programming, behavior-driven development, and user-focused design
- Refactored dispute management system components, reducing processing times by 40% and increasing user satisfaction by 25%
Technologies:
SmatSocial | Hyderabad, India
- Integrated Twitter, Facebook, LinkedIn, YouTube & Google+ APIs to be used as connectors in the platform
- Developed an extensive module to edit and detect image similarities using computer vision techniques
- Optimized backend to reduce load times with no tradeoffs by developing algorithms that could detect feed similarities both in text and images
- Lead a team of three to orchestrate a Chrome Extension to be used as a website assistant
- Developed a News Aggregator Application to compile news from various sources
- Developed an algorithm to rank person's social impact ranking based on their online presence
- Wrote complex webscrapers to get million+ text samples to be trained on AI engine
- Developed a robust module to store & maintain user info by integrating AWS S3 APIs into the system
Technologies:
George Mason University | Fairfax, VA
- GTA for IT Problem-Solving Using Computer Programming (IT 106) and Object Oriented Java Programming (IT 206) courses
- Summarized concepts, conducted labs and formulated questions for assignments
- Mentored over 100 students in programming fundamentals and object-oriented concepts
Technologies:
Featured Work
Projects that make a difference
Planwright.io
- Developed an AI-powered development platform that transforms how teams build software.
- Created AI mentorship system to guide developers and improve code quality in real-time.
- Built task orchestration features to streamline workflows and prevent technical debt.
Remove-BG.io
- Built a free, instant AI-powered background removal tool - no signup required, 100% free!
- Handles people, products, animals, cars, and graphics with full HD resolution output.
- Implemented features for transparent backgrounds, background replacement, and professional effects.
BitEdit.io
- Built a platform for streamlined developer workflows, simplifying tool setup.
- Integrated AI-based content creation features to enhance user productivity.
Conversational AI Assistant Platform
- Created a real-time chatbot for customer support and knowledge retrieval.
- Used LangChain, Falcon 40B, and the Azure OpenAI API with Pinecone for vector search and Redis for session management.
- Leveraged CrewAI and Langraph to coordinate multi-agent dialogues.
DeepRealty Advisor
- Built an AI-based property search assistant to automate manual research tasks.
- Used Playwright for browser automation and data validation.
- Adopted CrewAI and Langraph for user preference analysis and recommendation orchestration.
Optical Character Recognition (OCR) for WikiSource
- Developed a Django based Web Version for WikiSource to OCR PDF or DJVU files using Google Drive OCR API to update the relevant page in the WikiSource with the text
Skills & Expertise
AI & ML
Frameworks & Tools
Languages & Databases
Cloud & DevOps
Testing & Tools
Methodologies
Connect With Me
Contribution Activity
Loading GitHub contributions...