AI/ML Engineer · Computer Vision · Edge AI · IoT
Now Building
Production computer vision and RAG-powered AI systems for real-world operations.
Engineer mindset with product execution: solving practical problems through scalable AI and robust software architecture.
I'm Shubham Kulkarni — an AI/ML Engineer building production-grade computer vision systems that process 50,000+ vehicles daily on Indian expressways. Currently at Arya Omnitalk, I architect real-time detection and tracking pipelines using YOLOv8, DeepSORT, and TensorRT on edge devices (Jetson Xavier), achieving 92% mAP at 10ms inference latency.
My engineering DNA bridges hardware and software — from designing PIC microcontrollers for IoT sensor networks to building scalable Django/React web platforms. I won AIR 1 at Smart India Hackathon 2023 and hold internship experience at IBM (NLP/AI) and MathWorks (MATLAB/ML). I specialize in taking AI models from Jupyter notebooks to production — optimizing inference, building data pipelines, and deploying at scale.
I'm passionate about solving complex real-world problems through AI, and I'm actively seeking opportunities at product-based companies where I can build impactful, scalable AI systems.
My professional journey from internships to production AI engineering, with measurable business impact across computer vision, NLP, and embedded systems.
Building production-grade AI/ML systems for highway traffic management — vehicle detection processing 50,000+ vehicles/day with TensorRT-optimized inference at 12ms latency.
Engineered a multi-mode state machine for PIC microcontrollers with 5 operational modes, improving mode-switching accuracy by 20%.
Built an AI-powered Mental Health Tracker with NLP sentiment analysis achieving 90% accuracy in mood detection and <100ms API latency.
End-to-end AI workflow training in MATLAB — data preprocessing, model training, and evaluation. Earned 6 MathWorks certifications in ML, Deep Learning, and Image Processing.
Building AI/ML systems for highway traffic management — 50,000+ vehicles/day with 12ms latency.
Built multi-mode state machine for PIC microcontrollers, improving accuracy by 20%.
Built AI Mental Health Tracker with 90% accuracy mood detection and <100ms API latency.
End-to-end AI in MATLAB. Earned 6 certifications in ML, Deep Learning, and Image Processing.
A practical view of what I can ship end-to-end, from model training to production deployment.
Balanced across AI engineering, backend systems, cloud operations, and embedded intelligence.
A mix of production deployments, hackathon-winning systems, and full-stack products engineered for measurable outcomes.
Production-ready Document AI + RAG platform with 40+ FastAPI endpoints, hybrid retrieval (FAISS + BM25 + RRF), streaming Q&A, and analytics.
IoT-powered irrigation with AI-driven water optimization achieving 30% water savings. Raspberry Pi + soil sensors + weather APIs. AIR 1 at Smart India Hackathon 2023.
Autonomous robot with PID control, Arduino Mega, and 5-channel IR sensor array. Real-time path detection with interrupt-driven motor control.
Strong electronics foundation with a practical focus on intelligent systems, embedded computing, and applied AI development.
A collection of my professional certifications and credentials.
Focused on AI, machine learning, APIs, and deployment-ready engineering practices.
Feel free to reach out for collaborations, opportunities, or just to connect!