Portfolio & Applied Research
Work & Technical Systems
Spanning techno-economic models (Markets), early-stage venture diligence and commercialization (Startups), and applied physics, remote sensing, and robotics (Frontier Engineering).
Dictate: Local-First macOS Dictation & Context-Aware Speech Engine
A local-first, context-aware speech-to-text dictation application for macOS running speech recognition and formatting entirely on-device via whisper.cpp. Users hold a global hotkey, speak naturally, and cleanly structured text appears directly at their active cursor without audio frames ever leaving their computer.
Methodology & Architecture
Accelerated by Apple Silicon Metal GPU and the Apple Neural Engine to achieve sub-150ms transcription latency. Implements active frontmost window context detection, hesitation filler removal, and an on-device phonetic token biasing dictionary for specialized engineering terminology and colleague names.
Key Finding & Impact
Achieves sub-150ms latency with <35 MB idle RAM footprint on Apple Silicon, eliminating $15–$30/month cloud subscription fees and latency while guaranteeing complete air-gapped privacy in airplane mode.
Plot Passport: Satellite EUDR Verification Engine
A plot-level satellite verification system designed to help agricultural commodity exporters prove deforestation-free provenance under the strict EU Deforestation Regulation (EUDR) cutoff date.
Methodology & Architecture
Combines AlphaEarth geospatial foundation model embeddings with Hansen Global Forest Change layers and JRC forest masks, orchestrating multi-spectral temporal reasoning to eliminate canopy noise.
Key Finding & Impact
Achieved 1.00 recall vs 0.23 recall against standard FAO Whisp baselines on dense tropical smallholder plots.
Fuel Cell Vehicles: Techno-Economic & Infrastructure Feasibility
A quantitative review comparing proton-exchange membrane fuel cell vehicles (PEMFCs) against battery electric vehicles (BEVs) and internal combustion engines across lifecycle total cost of ownership.
Methodology & Architecture
Multi-variable sensitivity analysis across hydrogen production pathways, tube trailer logistics, compression losses, and station dispensing economics.
Key Finding & Impact
Demonstrated that fuel cell commercial viability is constrained by midstream distribution and dispensing Capex (over 58% of fuel cost), not stack efficiency.
Frontier Tech & Deep-Tech Venture Diligence Framework
Screened and advised 20+ early-stage startups across battery chemistries, EV charging, climate tech, and industrial robotics—establishing evaluation heuristics that separate coin-cell lab breakthroughs from commercial manufacturing yield.
Methodology & Architecture
Constructed a dual-axis diligence matrix testing fundamental scientific physics risk against manufacturing engineering friction and non-subsidized unit cost parity.
Key Finding & Impact
Conducted deep technical diligence on 9 ventures and recommended 3 for cohort admission; guided founders on aligning deep-tech IP with corporate customer willingness-to-pay.
National EV Adoption & GHG Abatement Scenario Modeling
Quantitative modeling of adoption trajectories for 2-wheelers, 3-wheelers, and commercial fleets under India’s Faster Adoption and Manufacturing of Electric Vehicles (FAME) scheme.
Methodology & Architecture
Modeled consumer payback thresholds, charging infrastructure density, and power-grid emissions intensity to calculate net lifecycle GHG reductions.
Key Finding & Impact
Quantified national GHG reduction curves to inform strategic policy positioning and automaker portfolio allocation.
Critical Success Factors for AGV Adoption in Manufacturing
Investigation of organizational, economic, and technological factors that determine the adoption of Automated Guided Vehicles (AGVs) in industrial warehouses and automotive assembly lines.
Methodology & Architecture
Applied the Fuzzy TOPSIS multi-criteria decision-making method across expert practitioners to rank investment priorities and operational hurdles.
Key Finding & Impact
Revealed that flexible floor reconfiguration and software integration outweigh upfront hardware cost in capital payback.
Programmable Selective Sorting System for Industrial Feeders
Design and prototyping of an intelligent optical sorting attachment integrated directly onto vibratory bowl feeders, eliminating secondary sorting conveyors in factory assembly lines.
Methodology & Architecture
Prototyped high-frequency pneumatic rejection gates coupled with optical sensors to inspect part orientation and geometry in real-time.
Key Finding & Impact
Eliminated footprint requirements of auxiliary sorting conveyors while maintaining continuous feeder throughput.
Counter-Drone Interdiction & Laser Defense Technologies
Survey of directed energy systems, RF jamming, and protocol spoofing countermeasures for neutralizing autonomous unmanned aerial systems (UAS).
Methodology & Architecture
Evaluated atmospheric attenuation limits for high-energy lasers, tracking precision, and electronic warfare countermeasures against swarm dynamics.
Key Finding & Impact
Structured technical comparison of soft-kill (RF/GNSS jamming) vs hard-kill (fiber laser) interdiction envelopes.