Hello! I am Suriya - Nice to meet you
I am currenly working on Data Science, and ML problems pertaining to Maps @Grab - such as Travel Time Estimation, Real-time traffic, Network Smoothening. The models which I have built are currenlty handling requests in order of Millions each day!
Prior to this, I was working in Banglore as a Software Engineer in Deep Learning where I built and deployed a small robo car with abilities to self drive, detect obstabls and signals - using Raspberry Pi, ROS, Flask and Pi-Camera
Some Topics which I am currenlty excited about
Working with Large Scale datasets pertaining to Maps, real-time traffic, travel time estimation, Speed prediction and network related smoothening. Also exploring building tools for explainable predicitons, monitoring model performance, feature drifts and automation pipeline using Kubeflow/ML-Flow
Built models for When-will-you-arrive ETA predictions for Cabs, deployed regionally over 100+ cities in South-East-Asia. Incorporated driver behaviour patterns, historic traffic conditions and built scalable data pipelines to ensure high quality data for model training and production deployment.
Built prototype for Self-Drive robot racing car (hotweels) and deployed using Flask, Raspberry Pi and Tensorflow. This uses Learnable activations, combined with NVIDIA proposed architecture for self-drive vehicles - Integration with ROS (Robot Operating System) to enable communications with other devices.
Built and locally served (flask) in-house tools for automation of business process which now intergrated with SCIO code base for business utility. Supported with Documentation, Database authentication, Customizable output templates and reduces human effort from ~2hrs to less than 5-sec.
Machine Learning, Digital Image Processing, Mining of Massive Datasets, Distributed Computing and Operating Systems
Using variants for Shallow/deep Neural Networks for learning latent semantic representaions for Graphs - with focus on obtaining Low-Dimensional Embeddings for trajectory clustering, Youtube users Graph and Protein-Interaction. Thesis provided upon request
Handled Lab courses for M.Tech and M.Sc on Machine Learning and Data Science. Built a course lab management system for assignment submission and grading for the couses I was handling.
Numerical Linear Algebra, Statistics, Functional Analysis and Measure Theory
Using Traditional ML and Image processing techniques to detect landmark points, construct various facical expression from EigenFace and its principle components. This method can generate expressions such as smiling, frowning for a given face which are not initally part of the training corpus. Thesis provided upon request
Calculus, Linear Algebra, Numerical Analysis, Graph Theory, Discreet Mathematics and Optimization Theory,
Implemented a simple file and directory transfer over LAN (server/client architecture) using socker programming in C (Unix Systems)
Apart from being a Data Science enthusiast, I enjoy most of my time being outdoors. I like trekking, reading harry potter, traveling - especially to beaches and sea.
When forced indoors, I try cooking (95th percentile says edible), photography and playing Asphalt 8