JAYADEEP GOWDAJAYADEEP GOWDA
AI & ML Developer | Robotics | Full-Stack Engineer
Engineering at the intersection of AI & Machine Learning, Robotics, and Full-Stack Project Building to architect high-impact, intelligent solutions.
About Me
As a Robotics and AI/ML Engineer, I specialize in bridging the gap between advanced Machine Learning architectures and physical autonomous systems. I drive innovation by fusing cutting-edge technology with an entrepreneurial mindset to architect scalable, high-impact intelligent solutions—from computer vision pipelines to localized LLM integration. My approach combines robotic precision with modern full-stack development to solve complex real-world challenges with technical excellence.
MY SKILLS
The Build Manifesto
Build Records
Curated Work
MEMO — Embodied Edge AI Companion
A production-grade embodied AI product architected from bare-metal firmware to high-level cognitive stack. MEMO represents a breakthrough in edge-native robotics, orchestrating a complex multi-modal system that runs 6+ concurrent AI models (Vision, Voice, Touch, LLM, Navigation) on a single Raspberry Pi 5. The custom Event-Driven Nervous System bridges the gap between cognitive reasoning and millisecond-level physical reflexes.
- Production-Grade Architecture: Fault-tolerant, thread-safe EventBus handling high-frequency asynchronous inputs without race conditions.
- Advanced AI Orchestration: Hybrid auditory stack (Vosk + Gemini 1.5 Flash) and parallel vision pipelines with <200ms end-to-end latency.
- Hardware Optimization: Kernel-level optimizations, including adaptive frame skipping and thermal burst management, sustaining 99.9% uptime.
- Full-Stack Ownership: End-to-end development of custom PCB design, 3D-printed chassis, firmware, and the React-based emotion engine.
Constructed a novel embodied intelligence architecture, optimizing LLM inference on edge hardware.
Cognis — Cognitive Pattern Engine
Designed Cognis to transform past reasoning into a first-class signal. Unlike stateless LLMs that allow repetitive cognitive loops, Cognis stores reflections as structured vector memory, detects recurring failure patterns in O(n) time, and intervenes to prevent repeated mistakes—shifting AI from just answering questions to actively preventing failure.
- Memory-Augmented Architecture: Architected a system to detect and store recurring reasoning failures across sessions.
- Temporal Pattern Scoring: Developed a custom algorithm driving 40–60% reduction in unproductive AI queries.
- Local-First Performance: Deployed a localized inference service achieving sub-200ms retrieval latency using Ollama.
Cognis solves the core flaw of conversational AI: LLMs answer, but they don’t remember when answering didn’t work.
YoloMart | Smart Retail Ecosystem
YoloMart unifies the fractured retail experience into a seamless 'Phygital' ecosystem. Unlike traditional stores with blind spots or disconnected apps, YoloMart treats every physical interaction—pickup, scan, cart add—as a digital event, eliminating checkout lines and empowering data-driven decisions.
- In-Store Digitization: RFID & WebSocket sync updates the digital cart instantly when physical items are handled.
- Frictionless Checkout: Scan & Go architecture eliminates queues with client-side comparison and instant payment.
- Context-Aware Assistant: AI analyzes cart contents in real-time to offer dietary insights and recipe synergy.
YoloMart bridges the gap between brick-and-mortar and e-commerce: picking up an item in-store is as analyzable as a click online.
Professional Journey & Achievements
A timeline of technical expertise, leadership roles, and recognition.
RnD Intern
Intern
Core Organizer
Technical Event Moderator
Contact Information
"I am always exploring new frontiers in tech. If you see a synergy or have an opportunity in mind, I invite you to review my detailed professional history."
Phone: +91 8310491224
Email: jayadeepgowda24@gmail.com
Location: Bangalore, India