Applied AI Research Background

Applied-AI Research Initiative

AI strategy meets
program delivery.
Applied, Generative &
Agentic AI.

MettleSense is an independent applied-AI research and product innovation initiative created by Rajeev Koul. The initiative serves as a platform for independent research, applied engineering, prototyping and development of working AI systems and product concepts across healthcare intelligence, human performance and autonomous enterprise workflows.

Executive Leadership & Scale

Executive Profile &
Global Track Record

Rajeev is a B.Tech, Anthropic Certified, CSM®, Six Sigma® & PMP® certified professional with 20+ years of global technology leadership experience across Fortune 500 companies and international organizations, driving AI (Applied, Generative and Agentic), Delivery Management, Program / Project and Product Management for software applications and large-scale implementations. Experienced in leading global engagements across India, US, UK, Philippines, Singapore, Malaysia, South Korea, Germany, Bangladesh and Sri Lanka.

Enterprise Program Delivery

Proven history of driving program management, technology strategy, and service delivery at scale across major multinational organizations including Samsung, Barclays, Citi, and Infosys.

AI Research & Product Innovation

Combines structured delivery frameworks with hands-on AI research and product development, translating advances in Generative AI, Agentic AI and machine learning into practical prototypes and production-oriented systems.

Achievements

Applied AI Engineering &
Research Accomplishments

Beginning in August 2025, Rajeev Koul undertook a structured professional-development sabbatical focused on applied AI research, architecture and product innovation. Building on more than 20 years of technology leadership and recent enterprise AI experience, he moved deeper into hands-on engineering—designing and developing multiple working AI systems, with the following three serving as flagship projects.

Cardiac Signal Processing

01. Project Shield — Edge Electrophysiology Engine

Architected and developed an AI-assisted ECG screening platform designed to identify patterns associated with nine cardiac rhythm and ECG abnormalities from single-lead wearable ECGs, later extending it to 12-lead interpretation using a deterministic PyTorch model achieving 0.902 mAUROC on the PTB-XL clinical dataset. Project Shield is now under the active mentorship and advisory of Dr. Iqbal H. Jaffer, MBBS, PhD — Cardiothoracic Surgeon, Assistant Professor of Surgery, and Program Director of the Cardiac Surgery Residency Program at McMaster University, Canada.

  • Ported the clinical 5-stage Pan-Tompkins R-peak detector algorithm (1985) to JavaScript.
  • Integrated Sample Entropy (SampEn) non-linear chaos analysis (m=2, r=0.2·SDNN) to trace irregular AFib patterns.
  • Constructed a mandatory 9-condition electrophysiology classifier that automatically corrects heart rate anomalies and suppresses false alarms.
Sports Science & Linguistics

02. Project Lean — Metabolic Forecasting Engine

Developed a body-intelligence analytics engine translating unstructured nutrition, training and physiological data into evidence-informed metabolic and recovery insights.

  • Built a fluid mass simulator modeling muscle glycogen repletion and sodium-induced water retention (Olsson & Saltin, 1970).
  • Created a high-speed fuzzy phonetic tokenizer (Double Metaphone + Levenshtein) supporting 220+ food aliases.
  • Mapped and resolved inputs natively across composite databases: Indian Food Composition Tables (IFCT 2017) and USDA FoodData Central.
Agentic Architecture

03. Project Balmukund — Autonomous Personal Operations

Architected a self-running, unattended email assistant running entirely on scheduled serverless triggers.

  • Designed a multi-agent routing loop utilizing Gemini and Claude models with self-correcting validation layers.
  • Created an idempotent financial parser tracking expenses, bank deposits, and upcoming subscription dues.
  • Integrated calendar scheduling systems automatically mapping dates and timezones (IST/UTC) with zero duplicate actions.
Rajeev Koul

Rajeev Koul

AI & Technology Leader | GenAI, Agentic AI & AI Product Development | Healthcare & Enterprise Automation | Delivery, Program & Product Leadership

Creator of MettleSense. Currently open to senior full-time leadership opportunities.

research@mettlesense.com