Architecting Edge AI & Health-Tech Platforms
Technology & Product Leader • Edge AI, LLM & SLM On-Device Deployment • TinyML & Wearable Diagnostics
Technology & Product Leader with 15+ years of experience across AI-enabled healthcare, wearables, embedded and connected-device platforms, with deep expertise in Edge AI, on-device SLMs/LLMs, TinyML, computer vision, and cloud-edge architectures. 6 years of hands-on experience deploying quantized neural networks and SLMs on constrained edge environments (Wear OS, iOS/Android, NNAPI via Google LiteRT). WIPO patent holder in AI-based activity classification.
15+ Yrs
EXPERIENCE
6 Yrs
EDGE AI & SLM
WIPO
PATENT HOLDER
+30%
ENGAGEMENT GAIN

Tushar Patil
Edge AI Systems Architect
Navi Mumbai, India
Sub-512MB SLM
LiteRT & Wear OS
WIPO Patent Holder
AI Activity Classifier
Executive Profile
Pioneering Edge AI & Intelligent Medical Hardware
Technology & Product Leader with 15+ years of experience across AI-enabled healthcare, wearables, embedded and connected-device platforms, with deep expertise in Edge AI, on-device SLMs/LLMs, TinyML, computer vision, and cloud-edge architectures. 6 years of hands-on experience deploying quantized neural networks and SLMs on constrained edge environments (Wear OS, iOS/Android, NNAPI via Google LiteRT). WIPO patent holder in AI-based activity classification.
Throughout my career as Chief Technology Officer and Lead Capability Architect, I have architected and deployed solutions spanning the complete spectrum of hardware-software co-design: from custom biomedical analog front-ends and low-power BLE firmware, to on-device neural execution (Google LiteRT) and cloud-native digital twin platforms on Google Vertex AI.
▹ Edge LLM & SLM (LiteRT & PyTorch)
▹ Computer Vision (32 Joint Tracking)
▹ TinyML & Multimodal Biosensing
▹ FDA Design Controls & ISO 13485
▹ Google Vertex AI & AWS Cloud
▹ Global ODM/CDMO Manufacturing
6 Years Edge LLM & SLM
Proven deployment of quantized Small Language Models on Wear OS and mobile under <340MB RAM.
WIPO Patent Holder
Patented activity classification machine learning algorithms from accelerometer and motion sensors.
Regulated Medical Devices
Deep mastery of FDA design controls, ISO 13485, IEC 60601, and EMC testing for therapeutic systems.
0-to-1 Product Velocity
Former CTO scaling connected health hardware from prototype to global mass production with tier-1 ODM/CDMOs.
Core Competencies
Edge AI Systems & Technical Pillars
Deep technical leadership unifying on-device neural inference, embedded firmware, biomedical hardware, and cloud intelligence.
Edge AI & On-Device SLMs
Deployment of quantized Small Language Models (SLMs) and LLMs (<512MB RAM) on mobile and Wear OS chips using Google LiteRT and PyTorch with Android NNAPI acceleration.
Computer Vision on the Edge
Low-latency on-device computer vision, including 32 anatomical body joints pose-estimation for real-time physiotherapy form correction and interactive AI co-pilots.
TinyML & Multimodal Biosensors
Microcontroller TinyML for continuous PPG, ECG, SpO2, and CGM biosignal processing, 1D Conformer feature encoders, and WIPO-patented AI activity classification.
Cloud-Edge Hybrid Architectures
Bridging constrained edge environments with Google Vertex AI and AWS pipelines, high-throughput BLE/NFC data ingestion, and personalized digital health digital twins.
Medical Devices & SaMD / SiMD
Full lifecycle engineering in regulated environments: FDA design controls, ISO 13485, IEC 60601, EMC compliance, and verification & validation (V&V) protocols.
0-to-1 Product Leadership
Transforming cutting-edge AI capability into mass-market hardware. Proven CTO track record scaling startups, BOM streamlining, COGS reduction, and global CDMO partnerships.
Key Product & Edge AI Initiatives
From open-sourced on-device Small Language Models (SLMs) and TinyML wearables to regulated digital therapeutics and clinical diagnostic analyzers.
Sub-512MB Multimodal Edge Cardiology AI Model (Cardiac SLM)
Edge AI & Small Language Models
Architected and open-sourced cardiac SLM, an edge-native multimodal cardiology AI model designed specifically for Wear OS smartwatches and mobile devices operating under strict constraints (<340MB RAM).
- •Built a lightweight 1D Conformer encoder processing 25Hz PPG signals to classify 5 cardiac rhythms with 97% accuracy and <10ms inference latency.
- •Integrated biosignals via cross-attention into a 4-bit student LLM with on-device clinical RAG (1,500 clinical guidelines).
- •Optimized for Google LiteRT (Android NNAPI) for real-time, ultra-low-power on-device cardiac triage without cloud dependency.
- •Published and open-sourced to the community on Hugging Face LiteRT Community.
AI-Based Human Metabolism Monitoring & Recommendation Platform
Edge AI & Health Platforms
Owned product strategy, architecture, and delivery for an AI-driven human metabolism monitoring platform combining edge-processed biosignals on mobile devices with cloud analytics.
- •Directed integration of multi-source biosignal data (ECG, SpO2, Continuous Glucose Monitoring - CGM) over low-power BLE/NFC pipelines.
- •Deployed Edge AI algorithms directly on mobile apps for real-time anomaly detection, filtering, and latency-free patient insights.
- •Integrated Google Vertex AI cloud pipelines for deep longitudinal metabolic digital twin modeling and personalized clinical recommendations.
AI-Based Exercise Analysis & Form-Correction Platform
Computer Vision & Edge AI
Directed product and engineering delivery for a real-time computer vision physiotherapy platform shipping an early on-device AI co-pilot product experience for rehabilitation.
- •Delivered real-time computer vision tracking of 32 anatomical body joints across iOS and Android using Google LiteRT.
- •Created an early AI co-pilot experience providing immediate voice and visual feedback on biomechanical form with zero cloud lag.
- •Tailored pose-estimation models and clinical form-scoring algorithms to assist physical therapists and patients.
Medical-Grade Wearable ECG & SpO2 Measurement Device
TinyML & Wearables
Owned product design and go-to-market for a medical-grade TinyML-capable wearable ECG/SpO2 device across chest bands and smartwatches.
- •Designed and deployed TinyML inference models on resource-constrained microcontrollers for real-time arrhythmia filtering.
- •Engineered secure BLE transmission pipelines to companion mobile applications for clinical review and cloud telemetry.
- •Navigated regulatory-grade biosensing standards for commercial deployment across Indian and US markets.
Design & Development of Wearable Health Monitoring Devices
Wearable Hardware & Ecosystem
Designed and developed wearable smartwatches and heart rate monitors for continuous health tracking, leading the complete product lifecycle from concept to mass production.
- •Led entire product lifecycle including ideation, industrial design, electronic hardware, firmware, and companion mobile apps.
- •Directed ODM manufacturing partnerships and international supply chains for high-volume consumer and healthcare scaling.
- •Integrated continuous optical heart-rate monitoring (PPG) algorithms and low-power power management systems.
COGS Optimization of Digital Therapeutic Device
Medical Devices & Therapeutics
Led a cross-functional redesign to optimize manufacturing cost while preserving clinical efficacy and performance for Edge AI driven prescription medical devices.
- •Streamlined Bill of Materials (BOM) and coordinated global component sourcing to reduce lead times and supply risks.
- •Established scalable mass manufacturing in partnership with the world's largest medical device manufacturer.
- •Delivered significant cost savings while maintaining strict device quality, biocompatibility, and tolerance standards.
Digital Therapeutic SaMD & SiMD Systems Management & Compliance
Regulatory & Quality Systems
Owned end-to-end product lifecycle engineering and technical management for digital therapeutic SaMD and SiMD devices across strict regulatory boundaries.
- •Led Verification and Validation (V&V), Electromagnetic Compatibility (EMC) testing, and comprehensive risk analyses.
- •Achieved FDA design controls readiness in full compliance with ISO 13485 and IEC 60601 medical device standards.
- •Aligned software architecture, embedded firmware, and clinical documentation for audit-proof submissions.
Human Body Composition Analyzer (Bioimpedance)
Diagnostic Instrumentation
Led end-to-end development of a multi-frequency bioimpedance body composition analyzer in collaboration with CDMO partners in China.
- •Engineered precision bioimpedance spectroscopy hardware, analog front-end signal conditioning, and calibration firmware.
- •Partnered with CDMOs in China to scale manufacturing quality, optimize injection molding, and refine PCB assembly.
- •Integrated companion smartphone application for body fat, muscle mass, and visceral fat trend analytics.
5-Part Blood Cell Counter (Hematology)
Diagnostic Instrumentation
Delivered an embedded Linux-based 5-part hematology blood cell counter from hardware architecture and firmware through manufacturing readiness.
- •Engineered high-reliability embedded Linux software and signal processing calibration for microfluidic cell sensing.
- •Integrated early cloud connectivity for centralized laboratory telemetry, automated reporting, and QC management.
- •Completed full design verification and clinical laboratory compliance transfer for commercial release.
Leadership Journey
Professional Experience
Lead Device Engineer — Capability Architect
Dr. Reddy's Laboratories
- Led scoping of AI integration into a multi-domain medical device roadmap, defining requirements and use cases with engineering and clinical stakeholders establishing a structured path for embedding AI capability into next-generation therapeutic devices.
- Led evaluation and architecture of AI/ML, GenAI and SLM integration for medical-device platforms, with focus on edge deployment, AI inference, hardware-software integration, and physical-device use cases.
- Led a cross-functional COGS optimization on a digital therapeutic device, streamlining the BOM and coordinating global sourcing to establish scalable manufacturing with the world's largest device manufacturer, reducing cost and lead time.
- Evaluated AI/ML vendors, platforms, and tools against product, technical, and compliance requirements for a regulated device environment, informing platform decisions for AI-enabled device initiatives.
- Drove AI-focused digital transformation initiatives across design, devices, and engineering functions, advancing the organization's AI and digital maturity alongside core device programs.
- Owned end-to-end product lifecycle management across concurrent therapeutic device programs including verification, validation (V&V), and EMC compliance delivering on schedule while maintaining regulatory readiness.
Chief Technology Officer
Oxstren Wearable Tech - Actofit
- Built an AI/ML-enabled wearable device and health diagnostics company from zero, owning technology and product strategy end-to-end vision, roadmap, architecture, and execution scaling core product lines from 0 to 1 across multiple device categories.
- Led development and integration of AI/ML, LLM and SLM capabilities across connected wearable and health-tech platforms, including AI-based solutions and deployment in edge environments like wearable smart watches, iOS, and Android mobile devices.
- Architected cloud-edge AI solutions, integrating AI models with physical devices, embedded systems, multimodal sensors, and real-time data pipelines.
- Architected 'Human Metabolic Health tracking & management platform' and deployed on Edge (mobile apps). Integrated wearable biosensors, embedded systems, cloud, and AI/ML (including Google Vertex AI pipelines) into unified product experiences, improving patient engagement by 30%.
- Directed product and engineering delivery for an AI-driven (Google LiteRT) mobile device-based exercise analysis platform, using computer vision tracking of 32 body joints across iOS/Android with real-time voice and visual feedback as an early AI co-pilot.
- Built and mentored cross-functional teams spanning hardware, embedded software, AI/ML, and systems integration, establishing a delivery cadence that sustained product velocity through multiple release cycles.
- Established OEM/ODM partnerships and global supplier networks, scaling products from prototype to mass production while aligning the technical roadmap with go-to-market objectives.
Assistant Professor & Research Associate
DJSCOE - Mumbai University
- Led early stage research and product discovery in BioMEMS and advanced instrumentation; mentored engineering teams and drove IP creation across multiple student-led product projects.
Embedded Application Engineer
AGD Biomedicals
- Engineered embedded hardware architecture and firmware for diagnostic products, improving performance, reliability, and time-to-market.
- Delivered an embedded Linux-based 5-part blood cell counter with integrated cloud connectivity as an early cloud-connected diagnostics product.
Technical Competencies
Skills & Domain Mastery
LLM & Edge AI Deployment
AI Data & Hardware Platforms
Cloud & Systems Architecture
Product Leadership & Strategy
Healthcare & Medical Devices
Delivery & IP Governance
Patents & Publications
Training of Classifiers for Identifying Activities Based on Motion Data
WIPO Patent WO2018134646
ML-based system for classifying user activity from motion data using trained classifiers, foundational IP underpinning AI-driven wearable products and real-time biomechanical analysis.
Education
M.Tech, Instrumentation & Control (Biomedical Instrumentation)
COEP Technological University, Pune (2011)
Grade: 8.35 / 10
Bachelor of Engineering, Biomedical Engineering
Mumbai University (2009)
Grade: 75.10%
Certifications
AWS Certified Cloud Practitioner
Amazon Web Services (AWS)
Regulatory Compliance: ISO 13485 & IEC 60601 (SaMD & SiMD)
Medical Device Quality & Safety Systems
HIPAA & Health Data Privacy Standards
Regulated Health-Tech Architecture
Get In Touch
Let’s Build the Next Edge AI Breakthrough
Open to executive leadership roles, advisory, and technical consulting across Edge AI systems, intelligent medical devices, on-device SLMs, and digital therapeutics.