Name: "Sambangi Narappagari Karthik"
Role: "Associate Software Engineer Trainee @ Transcend Street India"
Education: "B.Tech Information Technology — Andhra University College of Engineering"
Focus: ["AI/ML Systems", "Full Stack Engineering", "Cloud-Native Applications", "Multi-Agent Architectures"]
Philosophy: "Engineering products, not just code."I'm a software engineer with a strong foundation in AI/ML systems and full-stack product engineering, currently training at Transcend Street India. I design and ship multi-agent AI applications, scalable backend systems, and production-grade web platforms — with a product engineering mindset that prioritizes reliability, performance, and real-world usability over novelty.
My work spans LLM orchestration, cloud deployment (GCP, Vercel), RAG/agentic pipelines, and modern full-stack frameworks. I actively compete in national-level hackathons, building end-to-end AI products under tight timelines — from ML forecasting pipelines to AR-based assistive technology.
|
🎯 Open To
|
Languages
Frontend
Backend & Databases
Cloud, DevOps & Tooling
| Domain | Proficiency | Details |
|---|---|---|
| LLM Orchestration | ⭐⭐⭐⭐☆ | Multi-agent systems using Groq LLaMA 3.3 70B, agent-to-agent task routing |
| Generative AI APIs | ⭐⭐⭐⭐☆ | Gemini API integration for resume optimization & conversational support |
| Applied ML | ⭐⭐⭐⭐☆ | LightGBM, XGBoost, CatBoost ensembling with Optuna hyperparameter tuning |
| Computer Vision | ⭐⭐⭐☆☆ | YOLOv8-based object detection with OpenCV pipelines |
| NLP & OCR | ⭐⭐⭐☆☆ | Tesseract.js-based text extraction, accessibility-focused NLP tooling |
| AI System Design | ⭐⭐⭐⭐☆ | Orchestrator-agent architecture, SSE streaming, stateful task pipelines |
🧩 Synapse — Multi-Agent Productivity Assistant
A multi-agent AI productivity assistant featuring an Orchestrator agent that delegates to specialized Task, Calendar, and Email agents, built for the Google Cloud Gen AI Academy APAC 2026 Hackathon.
| Category | Details |
|---|---|
| Stack | Groq LLaMA 3.3 70B, FastAPI, SQLite, Server-Sent Events (SSE) |
| Scale | Multi-agent orchestration with real-time task delegation |
| Performance | Low-latency SSE streaming for live agent responses |
| Security | Isolated agent scopes, structured API request validation |
| Impact | Deployed live on Google Cloud Run; certified under Google Cloud Gen AI Academy APAC 2026 |
Synapse demonstrates a production-style orchestrator-agent pattern where a central controller routes user intents to domain-specific agents, each handling scoped responsibilities (tasks, scheduling, email) while maintaining shared context via SQLite-backed state.
📄 ATS Resume Optimizer
An AI-powered resume optimization platform that analyzes resumes against job descriptions and provides ATS-focused improvement suggestions using generative AI.
| Category | Details |
|---|---|
| Stack | Gemini API, React, FastAPI, Vercel |
| Scale | Handles multi-section resume parsing and JD-matching analysis |
| Performance | Real-time AI-generated feedback with responsive UI |
| Security | Client-side file handling with sanitized API payloads |
| Impact | Used as a core resume-tailoring tool across multiple job applications |
Built to bridge the gap between candidate resumes and ATS parsing systems, combining prompt-engineered Gemini API calls with a clean, recruiter-style UI for actionable resume feedback.
💬 Automatic Customer WhatsApp Support System
An AI-driven customer support automation system that handles WhatsApp conversations using generative AI for contextual, automated responses.
| Category | Details |
|---|---|
| Stack | Gemini API, Node.js, MongoDB |
| Scale | Automated conversational handling across customer support threads |
| Performance | Context-aware response generation with persistent conversation history |
| Security | MongoDB-backed session isolation per customer thread |
| Impact | Demonstrates scalable AI-driven customer support automation |
Designed to reduce manual support overhead by handling common customer queries automatically while preserving conversational context across sessions.
🚗 Automatic Number Plate Detection
A computer vision system for real-time vehicle number plate detection and logging.
| Category | Details |
|---|---|
| Stack | YOLOv8, OpenCV, MySQL |
| Scale | Real-time video frame processing for plate detection |
| Performance | Optimized inference pipeline for near real-time detection |
| Security | Structured MySQL logging of detected plate records |
| Impact | Applicable to smart parking, toll, and surveillance systems |
Combines YOLOv8 object detection with OpenCV preprocessing to accurately localize and log vehicle number plates, with detection results persisted to a relational database.
♿ ReadAR — WebAR Dyslexia Reading Assistant
An accessibility-focused WebAR application built for the Mercer | Mettl Visionary Hackathon 2.0, targeting the Healthcare theme for India's dyslexia demographic.
| Category | Details |
|---|---|
| Stack | Tesseract.js, OpenDyslexic Font, Web Speech API, Firebase |
| Scale | Real-time text capture, transformation, and speech-assisted reading |
| Performance | Low-latency OCR-to-speech pipeline |
| Security | Firebase-backed secure data handling |
| Impact | Assistive technology aimed at improving reading accessibility for dyslexic users |
ReadAR captures real-world text via camera, reformats it using dyslexia-friendly typography, and provides speech-assisted reading support — combining OCR, accessibility design, and AR principles.
📦 Flipkart Gridlock — Traffic Demand Forecasting
A machine learning competition entry focused on forecasting traffic/delivery demand using ensemble modeling techniques.
| Category | Details |
|---|---|
| Stack | LightGBM, XGBoost, CatBoost, Optuna, Ridge Stacking |
| Scale | Ensemble stacking across multiple gradient boosting models |
| Performance | Optuna-tuned hyperparameters for optimized forecast accuracy |
| Security | N/A (data science competition pipeline) |
| Impact | Built for competitive leaderboard ranking in Flipkart Gridlock Hackathon 2.0 |
A rigorous forecasting pipeline combining three gradient boosting frameworks with a Ridge regression meta-learner, tuned via Optuna for leaderboard-optimized performance.
Associate Software Engineer Trainee Transcend Street India Pvt Ltd · Hyderabad, India 2026 — Present
Currently undergoing structured technical training focused on software engineering fundamentals, professional communication, and enterprise development practices.
- Completed structured training modules covering software engineering fundamentals and workplace best practices
- Delivered technical RCA (Root Cause Analysis) presentations as part of soft-skills and professional communication training
- Handled enterprise workflows including structured documentation and process-driven communication
- Building foundational experience in enterprise-grade software development practices
Software Engineering Enterprise Training Technical Communication Process Documentation
Machine Learning Intern Polystack Technologies
Contributed to machine learning-focused projects, gaining hands-on experience in applied ML workflows.
- Worked on applied machine learning tasks within a professional engineering environment
- Gained exposure to real-world ML project workflows and collaborative development
Machine Learning Python Data Analysis
| Recognition | Details |
|---|---|
| Top 10% — B.Tech IT Department | Andhra University College of Engineering, Visakhapatnam |
| CGPA 8.92/10 | B.Tech Information Technology |
| Google Cloud Gen AI Academy APAC 2026 | Certificate of Completion — Multi-Agent AI System (Synapse) |
| Mercer | Mettl Visionary Hackathon 2.0 Participant | Built ReadAR — WebAR Dyslexia Reading Assistant |
| Flipkart Gridlock Hackathon 2.0 Participant | Traffic demand forecasting with ensemble ML models |
Google Cloud
Verified: certificate.hack2skill.com
current_focus:
learning:
- "Advanced DSA (LeetCode Medium/Hard)"
- "System Design Fundamentals"
- "Agentic AI Architectures"
building:
- "Production-grade multi-agent AI systems"
- "Full-stack AI-powered applications"
exploring:
- "RAG pipelines & vector databases"
- "Cloud-native AI deployment patterns"
open_to:
- "Software Engineering Roles (SDE / AMTS)"
- "AI/ML Engineering Opportunities"
- "Open Source Collaboration"
"Great engineering is invisible — it just works, at scale, under pressure."