Skip to content

jai2992/jai2992

Folders and files

NameName
Last commit message
Last commit date

Latest commit

Β 

History

6 Commits
Β 
Β 

Repository files navigation


πŸ‘¨β€πŸ’» About Me

I'm a B.E. Computer Science (AI & ML) student at Sri Eshwar College of Engineering (Expected May 2026, CGPA: 8.6), currently working as a Software Developer Intern at nBase2 Systems, Bangalore β€” building real-world AI systems at the edge.

I love the full stack of AI: from training models in Python to squeezing them onto microcontrollers in C++. Whether it's a RAG pipeline processing 20K+ pages or a mesh network built on ESP-NOW, I enjoy solving problems end-to-end.

  • πŸ”­ Currently working on audio ML models and edge inference on TI AM62A/AM62D SOCs
  • 🌱 Deep-diving into MultiAgent frameworks, TFLite optimization, and RAG pipelines
  • πŸ† Multiple national hackathon winner (β‚Ή65,000+ in prizes)
  • πŸ“ Based in Bangalore / Coimbatore, India

πŸ’Ό Work Experience

nBase2 Systems Pvt Ltd β€” Software Developer Intern Feb 2025 – Present

  • Developed audio ML models in Python; optimized TFLite models in C++ on TI AM62A/AM62D SOCs
  • Built end-to-end edge inference workflow: training β†’ quantization β†’ conversion β†’ firmware integration
  • Prototyped RAG-based document retrieval pipelines for a MultiAgent framework (20K+ page documents)
  • Developed & deployed a Leave Management System (50+ users) with Next.js, PostgreSQL, AWS Amplify & RDS

πŸš€ Projects

Python YOLO ESP-NOW Raspberry Pi ESP32

  • Event-driven edge AI framework β€” YOLO detection activates only on motion trigger, minimizing power consumption
  • Self-organizing multi-hop mesh network via ESP-NOW + Protocol Buffers β†’ 76% reduction in payload size

TensorFlow DenseNet121 Transfer Learning OpenCV Streamlit

  • Defect classifier on 805 images across 6 classes β€” 86% accuracy despite constrained dataset
  • Handled class imbalance with augmentation; deployed as an interactive Streamlit web app

Python TensorFlow Librosa MFCC

  • Spoken digit (0–9) classifier using MFCC + Mel Spectrogram features β€” 94% accuracy
  • Full audio ML pipeline: preprocessing β†’ feature extraction β†’ training β†’ evaluation

πŸ† Achievements

Competition Prize
πŸ₯‡ 1st Place JITHACK'24 β€” JIT Bangalore β‚Ή50,000
πŸ₯‡ 1st Place Hackerrupt'24 β€” GUVI @ IITM Research Park β‚Ή15,000
πŸ₯ˆ Runner-up AI Hackathon β€” Karpagam College of Technology β€”
πŸ₯ˆ Runner-up Prompt Engineering Competition β€” CIT Coimbatore β€”
πŸ₯‡ 1st Place Paper & Project Presentation β€” PA College of Engineering β€”

πŸ› οΈ Tech Stack

Languages Python C++ JavaScript SQL

ML / AI TensorFlow TFLite YOLO OpenCV Librosa

Edge / Embedded Raspberry Pi ESP32

Web / Frameworks Next.js React Node.js Streamlit

Tools & Cloud AWS Docker PostgreSQL Git


πŸ“œ Certifications

  • πŸŽ“ Supervised Machine Learning β€” DeepLearning.AI
  • ☁️ Generative AI & LLMs β€” Google Cloud
  • ☁️ AWS Cloud Practitioner β€” Amazon Web Services

"Ship real things. Learn from hardware. Build what doesn't exist yet."

About

Hello everyone this my profile

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

 
 
 

Contributors