Mihir Kapadia Email & Phone Number

Leiden, South Holland, Netherlands

Delft, South Holland, Netherlands

• Introduced new features to perClass Mira software for 3D (hyperspectral) data processing - calibration, noise reduction, preprocessing, dimensionality reduction.}• Analysed the feasibility of spectral unmixing, regression algorithms for post harvest food quality parameters for NIR and VNIR spectral ranges - Brix, firmness, weight, protein content.}• Created libraries for functions tailored to workflows, ensuring performance consistency on multiple datasets.• Conducted literature reviews for spectral processing algorithms for various applications

Delft, South Holland, Netherlands

• Leading and managing projects related to on-board computer and data handling systems.• Providing technical expertise in data handling architectures, components, and security.• Conducting research to enhance data processing, transfer, and security.• Collaborating with stakeholders and ensuring effective communication.• Overseeing the development and implementation of research outputs to be integrated in the system.

Delft, South Holland, Netherlands

• Conducted a comprehensive literature review to gather relevant information and identify state-of-the-art techniques in odometry and related fields.• Developed and implemented odometry algorithms using advanced techniques such as dynamic state estimation and Bayesian processing.• Engaged in requirement engineering and stakeholder discussions to understand project goals and constraints.• Utilized model-based system engineering (MBSE) principles to design and analyze complex systems.• Supported systems engineering design tradeoff activities by collaborating with cross-functional teams to identify tradeoffs and make informed decisions regarding system design.

Amsterdam, North Holland, Netherlands

Distributed & Interactive Systems GroupResearch Thesis - Few-Shot Emotion Recognition with intelligent voice assistants and wearables : Learning from few samples of speech and physiological signals.- Performed a literature review of state-of-the-art feature engineering and processing techniquesfor speech and physiological signals.- Created Multimodal Physiology-reinforced Siamese Network, a novel unsupervised learning algorithm to address challenges and constraints of emotion/sentiment classification from real-world speech for voice assistants (Few Shot Learning)- Improved state-of-the-art emotion recognition benchmarks by a margin of 8-12%.- Enhanced model performance for robustness against background noise from real-world within2-5% margins of baseline.Tech stack : Python, TensorFlow, PyTorch, Docker, Pandas, Scikit-learn, Librosa, OpenCVSupervisors : Abdallah El Ali, Alle-Jan van der Veen, Pablo Cesar

Delft, South Holland, Netherlands

Coach for TU Delft Introduction Programme - Guided and introduced international incoming students to TU Delft programmes.

Delft, South Holland, Netherlands

Responsible for guiding students on weekly tasks oriented in design of a product through Systems Engineering, grading of weekly group projects on outcomes and guiding ideas for brainstorming.

Delft, South Holland, Netherlands

Responsible for guiding students in assignments and grading and providing feedback to students on their weekly work. Collaborated on creating assignment and answering questions from students.

Mumbai Metropolitan Region

• Contributed to Software and Hardware Design, Modelling and for Computer Vision based BrokenRail & Intrusion Detection System for Indian Railways• Curated and managed the first dataset for rail anomalies obtained from sensing experiments inIndian railways, Benchmarked the severity of rail cracks based on several feature sets.• Tailored computer vision-based anomaly detection pipeline’s architecture to increase sensitivitytowards artefacts from Indian setting;

Mumbai, Maharashtra, India

• Responsible for Engineering, Verification & Validation of Train Control Systems for Indian Railways • Drafted on technical system requirement specifications, System architecture for various packages in ETCS and Interlocking projects in Indian Railways and Metros. • Contributed to the technical feasibility analysis for 3 digitalization pilot projects in India - Broken Rail & Intrusion Detection System, CT Mobile Telematics, EoTT.

Mumbai Metropolitan Region

• Established a data-oriented approach for the Mobility business in India with focus on fastest growing product and project businesses. Helped in improving product sales by close to 10%.

Master Of Science - Msc, Electrical Engineering - Signals & Systems

Signals and Systems Track. Selected Courses - 1. Signal Processing - - Statistical Digital Signal Processing and Modelling - Estimation.

Bachelor Of Technology - Btech, Electrical Engineering

Activities and Societies: Memberships - IEEE Student Member, Cutting Edge VisionariesSubjects - Control Engineering Courses - Control.

Indian School Certificate Examination - 2012, High School/Secondary Diploma Programs, 90/100

Science Major - Physics, Chemistry, Mathematics