13 years engineering intelligent systems — from real-time computer vision pipelines to enterprise-grade LLM applications that ship to production.
I am a Senior Lead Engineer with over 13 years of hands-on experience designing and deploying production AI systems. My expertise spans the full stack of modern AI — from low-level CUDA GPU programming and embedded edge inference to cloud-native LLM applications with retrieval-augmented generation.
Across my career I have built lane detection and autonomous driving systems, enterprise facial recognition platforms, IoT-powered toll management, and conversational AI chatbots that reduced customer query resolution time by 40%. I have a proven track record shipping models from experimentation to production on AWS, Azure, and GCP.
Currently deepening my work in Generative AI and LLM orchestration — building RAG pipelines, agentic systems (LangGraph, AutoGen), and multi-model deployments using LangChain, LlamaIndex, and AWS Bedrock. I believe the best AI engineering is invisible: it just works, reliably, at scale.
I thrive in cross-functional environments that bridge research and product — translating cutting-edge AI advances into tools real people use every day.
Real-time lane detection and tracking using deep learning and computer vision. Handles challenging scenarios including curved roads and partial occlusions with high accuracy at production-level throughput.
Deep learning system that detects vehicle lane drift and delivers visual and audio alerts in real time to prevent accidents. Built for reliability across diverse road and lighting conditions.
IoT-powered toll system using a custom-trained YOLOv4 model on Raspberry Pi 4 to detect and count unregistered vehicles. Integrated with Infinite Uptime for predictive maintenance and high-frequency fault diagnostics.
Enterprise ML framework automating workspace management: Attendance, Cafeteria, Visitor & Ticketing; HVAC and Lighting zone control; Desk and Meeting Room booking — all driven by a unified facial recognition backbone.
Regression-based predictive model for second-hand vehicle pricing using multi-source datasets (Quikr etc.). Analyses mileage, make, year, and condition across dozens of parameters to generate accurate market estimates.
Comprehensive AI health assistant powered by Gemini 1.5 Pro, RAG, and LangChain. Supports symptom diagnosis, treatment suggestions, and personalised preventive healthcare advice — enhancing patient care and medical consultations.
Enterprise chatbot powered by GPT-4o, LangChain, and AWS Bedrock. Intent recognition and dynamic response generation reduced query resolution time by 40% while measurably improving satisfaction scores.
Transformer-based summarization (Pegasus) for PDF and DOCX documents using semantic relevance extraction. Reduced manual review time by 60%, boosting productivity in legal and academic workflows.
Open to senior AI/ML engineering roles, consulting engagements, and research collaborations. Let's talk about how AI can solve your toughest challenges.