I'm an AI/ML Engineer and Full Stack Developer based in Bengaluru, available for remote work and freelance projects.

I don't just build models — I build products. That means taking an AI idea from raw data all the way to a deployed application that solves a real problem. I work across NLP, Computer Vision, automation pipelines, and full-stack web development, and I'm comfortable owning the entire stack.

What I'm best at:

• Taking a vague AI problem and turning it into something shippable
• Building end-to-end ML pipelines — data → model → API → UI
• Integrating AI (pose estimation, LLMs, RAG) into real-world applications
• Automating workflows that save businesses real time and money

💼 Available for: Remote AI/ML roles · Freelance AI product builds · NLP & Computer Vision consulting

A real-time coordination platform built for motorcycle groups — from the first meeting point to the last junction.

JourneySync solves the chaos of group riding: scattered WhatsApp threads, no live map sync, lost riders, and zero safety net. Built end-to-end as a cross-platform mobile app.

Flutter Firebase Google Maps SDK FastAPI WebSockets Supabase

Core Features:

• 🗺️ Live group tracking — real-time location sync across the entire ride group on a shared map
• 📲 QR-based lobby onboarding — scan once, join the group ride instantly; no manual sharing needed
• 🎙️ Pack intercom with wind noise reduction — in-ride voice communication optimized for highway speeds
• 🏔️ Offline topographic maps — full route visibility even in dead zones and remote terrain
• 🚨 Crash detection + SOS beacon — auto-detects falls via sensor fusion and triggers emergency alerts with GPS coordinates

Turns any smartphone into a clinical-grade cardiac screening tool. Trained on the PhysioNet 2016 dataset (3,240+ heart sound recordings) using mel-spectrogram preprocessing and deep learning, reaching 92.23% AUC — then compressed with TensorFlow Lite for real-time, offline, on-device inference.

Python TensorFlow/Keras TensorFlow Lite Librosa OpenCV Streamlit

• Deep learning on mel-spectrograms achieves 92.23% AUC for heart sound classification
• TensorFlow Lite optimization cuts model size by 91% (1.33MB → 0.12MB) for real-time mobile inference
• Streamlit web app with QR-based workflows for offline, mobile-first cardiac screening

NLP-driven recruitment platform that parses, ranks, and semantically matches resumes to job descriptions with explainable, bias-mitigated scoring — plus a proactive AI chatbot that gives candidates real-time resume feedback before they apply.

React.js Node.js Hugging Face spaCy Supabase Rasa

• Semantic JD-to-resume matching with explainable, bias-mitigated candidate scores
• Proactive AI chatbot delivers real-time resume feedback before submission
• Collaborative HR dashboard for evaluation and intelligent candidate insights

Real-time Gen-AI pipeline that detects, identifies, and tracks vehicles across citywide CCTV networks — restoring unclear footage and forecasting a target's next move to enable fast, intelligent interception.

SAM3 Qwen ImageEdit 2509 LLaMA 3.2 Real-ESRGAN Stable Diffusion CCTV Processing

• Segmentation-level vehicle isolation (SAM3) combined with clarity restoration via Real-ESRGAN + Stable Diffusion
• Route-forecasting engine predicts a target vehicle's next junction and ETA
• Real-time geo-coordinate alerting dashboard built for law enforcement automation

Languages

AI / ML

Frontend

Backend & Databases

Automation & Tools

Reach me: LinkedIn · Portfolio · YouTube