Data Scientist - Machine Learning (2-6 yrs)

Description :

The Center of Data for Public Good (CDPG) is an interdisciplinary research center at the Indian Institute of Science, Bangalore, focusing on Data Science for translational research in domains such as Traffic & Transportation, Air Quality, Geospatial Analytics, Agriculture, and Health Care.

We are building a Digital Twin for mobility by combining many data sources such ANPR, telco, AVLS, and incident data with key measurements such as vehicle counts and speeds extracted from video feeds from surveillance cameras.

These measurements will be used in building a mobility model which can predict traffic congestion and anomalies, and also support short-term and long-term scenario analysis for use cases such as road closures, major events and infrastructure planning. As part of this exciting project, we are looking for experienced Data Scientists with good experience in building Computer Vision applications.

Responsibilities :

- Develop vision/ai/ML techniques to detect and classify vehicular and other road objects, track them / re-identify them and obtain KPIs for traffic analytic, to be deployed at scale for about 6000 cameras.

- Building vision applications at scale that require optimizations such as pruning/architecting Yolo-like models and tuning them for efficiency in terms of VRAM usage, latency, model size, etc.

- Deploying models in large-scale inference systems like Triton and potentially optimization at CUDA level

- End-to-end MLOps with Experiments, model tuning, experiment tracking, architecture search and inference deployment

- Understand and implement latest research papers on topics related to this space

- Experience with YOLO and other object detection/recognition of models including their architecture, fine-tuning and deployment at scale

- Experience with TFlow, Torch, CUDA

- Experience with MLOps tools like MLFlow, KubeFlow, Ray, and Dask

- A good grasp on computer vision and ML fundamentals, linear algebra, probability and statistics.

- Ability to translate and innovate on top of state-of-the-art research papers in the CV/AI-domain into code

- Proficient in Python with experience in MultiProcessing, Greenlets, C/C++ Wrappers and asynchronous programming