Building intelligent software
from data to intelligent products.
I design full-stack systems that combine machine learning, analytics, and modern backend architecture.
Nexus
Traditional communication suites are fragmented (Slack vs. Notion vs. Miro vs. Zoom). I wanted to build an AI-native workspace where huddles, chat, docs, and whiteboards are unified under a single, real-time multiplayer context.
<50ms
Video Latency
1,080p
HD Resolution
100%
Semantic Search
Zero
Trust Access
System Architecture
An intelligent workspace combining real-time edge media with a multiplayer CRDT canvas and secure semantic indexes.
Edge Media Network
Offloads audio/video routing to Stream global edge networks, maintaining sub-50ms latency for huddles.
Semantic Search Index
Processes canvas edits into Pinecone vector storage via Liveblocks webhooks and Google Gemini embeddings.
Zero-Trust Boundaries
Serverless brokering using Firebase Admin SDK and Clerk tokens to guarantee cross-tenant file isolation.
Interactive Workspace Control Panel
To test the unified nature of Nexus, I built simulation triggers mapping the underlying backend actions. Switch tabs below to execute simulated server actions, checking how AI summarization, semantic querying, and zero-trust tokens respond in real-time.
Developer Toolkit
Apps, tools, and platforms built for learning and scale.
CricSphere
The Ultimate Cricket Destination with AI-ML Powered Analytics. A comprehensive hub for live scores, schedules, and player statistics with real-time prediction engines.
AetherAI
Built a real-time forecasting pipeline using XGBoost (94.2% confidence) and Gemini 1.5 Flash to translate predicted AQI into actionable, personalized health advice.
Engineering Highlights
Raw metrics from the datasets, models, and systems I've built. No fluff, just scale.
22,007
Matches Processed
638K+
PvP Records Analyzed
11K
Images Classified
80%
CNN Brand Accuracy
<360μs
Vector Math Latency
~1.1s
Token Revocation Delay
<50ms
WebRTC Edge Delay
<50ms
ML Inference Latency
Development Journey
A chronological timeline of systems built, algorithms mastered, and architectural decisions made.
Programming Foundations
Algorithmic reasoning & Core OOPs
Began with Java and Data Structures, establishing a strong foundation in core algorithms, complexity analysis, and database normalization.
Programming Foundations
Algorithmic reasoning & Core OOPs
Began with Java and Data Structures, establishing a strong foundation in core algorithms, complexity analysis, and database normalization.
Full-Stack & Applied AI
Modern web frameworks & NLP datasets curation
Mastered the MERN stack while contributing to voice dataset tools. Designed fully responsive user interfaces and robust REST endpoints.
Full-Stack & Applied AI
Modern web frameworks & NLP datasets curation
Mastered the MERN stack while contributing to voice dataset tools. Designed fully responsive user interfaces and robust REST endpoints.
Analytics & Caching Systems
Sports telemetry processing & model serving
Engineered CricSphere parsing engines for heavy JSON dataset workloads, integrating efficient in-memory caches to speed up dynamic predictions.
Analytics & Caching Systems
Sports telemetry processing & model serving
Engineered CricSphere parsing engines for heavy JSON dataset workloads, integrating efficient in-memory caches to speed up dynamic predictions.
Real-Time AI Collaboration
High-concurrency document sync & telemetry
Architecting Nexus, incorporating Liveblocks Yjs conflict-free document sync, zero-trust storage URL brokering, and Gemini embedding matching.
Real-Time AI Collaboration
High-concurrency document sync & telemetry
Architecting Nexus, incorporating Liveblocks Yjs conflict-free document sync, zero-trust storage URL brokering, and Gemini embedding matching.
Work Experience
My professional career internships, contributions, and industry highlights.
Deep Learning Intern
Central Institute of Tool Design (CITD)
- Built an EfficientNet-based CNN classifier trained on 11,000+ images across 50 classes.
- Achieved ~80% accuracy using transfer learning and data augmentation.
- Designed preprocessing pipelines and evaluation workflows for model optimization.
Deep Learning Intern
Central Institute of Tool Design (CITD)
- Built an EfficientNet-based CNN classifier trained on 11,000+ images across 50 classes.
- Achieved ~80% accuracy using transfer learning and data augmentation.
- Designed preprocessing pipelines and evaluation workflows for model optimization.
AI Creators Program Intern
Swecha.org
- Contributed to Telugu language AI initiatives through dataset collection and NLP workflows.
- Explored LLM, tokenization, and NLP model development concepts.
- Worked with Text-to-Speech and Voice Avatar technologies.
AI Creators Program Intern
Swecha.org
- Contributed to Telugu language AI initiatives through dataset collection and NLP workflows.
- Explored LLM, tokenization, and NLP model development concepts.
- Worked with Text-to-Speech and Voice Avatar technologies.
Proof of Work
Competitive programming dashboards, social proof, and cloud foundations.
Competitive Profiles
97
Top 3% Nationwide
Young Turks Percentile
834
253 Problems Solved
CodeChef Max Rating
113+
Data Structures & Algorithms
LeetCode Solved
69+
Problem Solving
GeeksforGeeks
Salesforce Agentblazer
Champion 2026
Cloud & AI Certifications
AI Upskilling Certificate: Technical Foundation
Qualcomm • Nov 2025
Oracle Cloud Infrastructure Foundations Associate
Oracle • Sep 2025
Artificial Intelligence Fundamentals
IBM SkillsBuild • Jul 2025
Generative AI for Data Science
Coursera • May 2025
AWS Academy Graduate - Data Engineering
AWS Training • Apr 2025