Thiru Kumaran

A Data Science professional specialized in data synthesis, data collection, statistical analysis and ML model development with a combined experience of over 6 years in scientific and commercial big datasets. Have played a key role – in translating business requirements to technical requirements; in giving actionable insights to internal and external stakeholders. From ideation to productization, led research projects and guided Junior Data Scientists. Key competencies include:

• Presentation to stakeholders (internal/external)

• Collaboration

• Critical thinking

• Project proposal

• Publishing research articles

• Solution architecture

Quadrant Global Pte Ltd. Jan 2022 – Jul 2022

Senior Data Scientist

Led Research and Development projects that gave actionable insights to the Clients, the Marketing and the Product team to strategize and improve the point-of-interest (POI) location data collection methods and accuracy. Method of work management: Agile

• Developed and deployed a trend forecast predictive model of time-series data – vis-à-vis hypotheses analysis.
• Developed and deployed statistical model to predict the relative drop in live POI data-feed.
• Developed smoothing Algorithm (PoC) for improving the POI location data accuracy (GPS/INS sensor fusion).
• Developed and deployed Outlier algorithms, as a means of Anomaly detection and User Behavior, that gave actionable insights to the Marketing team and the product team.
• Led a team of Junior Data Scientists that developed a recommendation system using Scoring metrics.
• Led a team of Junior Data Scientists that developed a NLP-classifier models for automating Quality Assurance process. 
• Handled Client complaints by carrying out Proof-of-Concept studies. 
• Prepared “Confluence” reports of all projects as a means of business communication.
• Reviewed codes (Github); evaluated algorithms.

National University of Singapore Oct 2016 – Dec 2021

Was responsible for aerospace related data collection through experiments and numerical simulations, data analysis and development of deep learning models. Published scientific research articles. 

Research Scientist, Temasek Laboratories at NUS (Singapore, Singapore) | Jan 2020 – Dec 2021

• Carried out aerospace experiments and numerical simulations for data collection.
• Developed statistical models to ascertain data population.
• Created training datasets and facilitated its storage in Supercomputing Clusters (NSCC, Singapore).
• Developed and deployed state-of-the-art Physics-Informed Neural Networks that reduced computation time by 10 folds.
• Developed and deployed Geometric programming optimization architecture that optimized the flight characteristics.
• Developed and deployed data-driven model, that reduced simulation time by 5 folds, to simulate the performance of combustion engine.
• Created Reports and Presentations as a means of business communication with internal and external stakeholders.
• Raised 2 seed grants.
• Published research articles in the prestigious AIAA Scitech Forum conferences.
• Led a team of undergrad and masters students in Computational Fluid Dynamics projects.

Research Fellow, Dept of Mechanical Engineering, NUS (Singapore, Singapore) | Oct 2016 – Dec 2019

• Developed Deep Reinforcement Learning model, a Proof-of-Concept, to increase flight performance.
• Developed and deployed ANN-coupled Bayesian Architecture for optimizing the aircraft wing shape.
• Parallelized and deployed a multi-batch edge detection algorithm for a 2TB unstructured dataset that reduced the usual image-by-image computational analysis time.
• Developed and deployed Multilayer Perceptron Neural Network for the prediction of flight performance metrics and achieved Optimal flight performance.
• Developed and deployed Geometric Programming Optimization Architecture as a means of System modeling.
• Collaborated with The University of Queensland’s Center for Hypersonics and DSO Singapore for Research and Development purposes.

Doctor of Philosophy, Mechanical Engineering, National University of Singapore

My doctoral dissertation was conducted in consultation with the then Assistant Prof. Estruch-Samper David and examined the influence of bio-inspired roughness structure called convergent riblet in incompressible boundary layer. During my research I carried out numerical simulation of partial differential equations using the scripting language Python. Apart from that, I have applied Machine Learning algorithms, ranging from Regression, classification to MLP LSTM-RNN, to experimental and numerical data.