Sabarish Vadarevu

AI Researcher

Bengaluru, Karnataka, India11 yrs 2 mos experience

Most Likely To SwitchAI ML Practitioner

Key Highlights

Led development of ML systems for AI platform.
Expert in computational fluid mechanics and turbulence modeling.
Strong background in data science and machine learning.

Stackforce AI infers this person is a Data Science and Machine Learning expert in the SaaS industry.

Contact

Skills

Core Skills

Machine LearningComputer VisionComputational Fluid MechanicsNumerical AnalysisModeling

Other Skills

Agile MethodologiesArtificial Intelligence (AI)BenchmarkingBusiness AnalysisBusiness AnalyticsC++CommunicationData AnalysisData ManagementData ScienceData VisualizationDeep LearningEmissions AnalysisFinancial AnalysisFinancial Modeling

About

Building the machine learning systems behind Akridata's low-code, data-centric AI platform for computer vision. The platform offers lifecycle management for data and models. This involves data pipelines and management, data labeling, generation, and curation, along with model training, optimization, analysis, and monitoring for concept and data drift. As the team manager and technical leader, I adapt and improve the state-of-the-art in CV/ML, drive priorities and systems design, collaborate with other engineering teams, and define processes for test-driven development and continuous integration. Formerly, I was a Research Fellow in Computational Fluid Mechanics, with an extensive background in numerical methods for non-linear ODEs and PDEs, optimization, calculus of variations, and matrix-free methods for sets of linear equations.

Experience

11 yrs 2 mos

Total Experience

2 yrs 9 mos

Average Tenure

6 yrs 11 mos

Current Experience

Akridata

3 roles

Senior Manager, Data Science

Promoted

Jan 2023 – Present · 3 yrs 5 mos

Tech Manager, Data Science

Feb 2022 – Jan 2023 · 11 mos

Tech Lead - Data Science

Jul 2019 – Feb 2022 · 2 yrs 7 mos

Responsible for an in-house machine learning library aimed at simplifying visualization, sampling, and annotation for data science teams on Akridata's data management and exploration platforms - including research, design, development, validation, and testing of ML techniques.
Training and benchmarking efficient models for computer vision (CV) focusing on the autonomous vehicle (AV) domain.

Machine LearningComputer VisionData ManagementVisualizationModel TrainingBenchmarking

University of melbourne

Research Fellow

Feb 2017 – May 2018 · 1 yr 3 mos · Melbourne, Australia

Understanding turbulence dynamics in terms of travelling coherent structures is important to our modelling efforts. Vortex-streak structures are suspected to be the building blocks of turbulence, but direct evidence for such structures in high Reynolds number flows is unavailable to date. The birth and growth of these structures are adequately described by linear processes (before their eventual decay and regeneration). Such linearity is explored in our work through the impulse response in high Reynolds number flows. Impulse response is a reasonable starting point, since "bursting" events that generate vortex-streak structures are spatially concentrated. Our numerical simulations confirm that the response contains vortex-streak structures that have some degree of self-similarity, paving the way to incorporating such self-similar coherent structures in further efforts to model and understand turbulence.

Computational Fluid MechanicsNumerical MethodsTurbulence DynamicsSimulationNumerical Analysis

University of southampton

2 roles

Tutor

Oct 2015 – Feb 2017 · 1 yr 4 mos · Southampton, United Kingdom

Helping foundation year students with worksheets in mechanical science.

Laboratory Demonstrator

Oct 2014 – Jan 2017 · 2 yrs 3 mos · Southampton, United Kingdom

Involves setting up and explaining the following experiments to groups of ~10 undergrads:
1) Measuring forces and moments on a symmetric airfoil (finite and infinite) at different AoA and flap angle in a 5ftx7ft wind tunnel through pressure taps on the surface, and comparing the results to those from sensor data and thin airfoil theory.
2) Measuring the velocity profile and thickness of laminar and turbulent boundary layers (zero pressure gradient) in a small wind tunnel (test section ~ 1ftx1ft), and comparing to analytical approximations.

Indian institute of technology, madras

Graduate Teaching Assistant

Aug 2012 – Apr 2013 · 8 mos · Chennai Area, India

For the semester of Aug-Nov 2012, I was one of two tutors for a group of 50 undergraduate students in their 2nd year. We helped the students solve problems in the basics of fluid mechanics. We had also graded their course-work and exams.
For the semester of Jan-Apr 2013, I was part of a 4-member assistant team responsible for tutoring a course in aircraft design (taken by about 80 students). The course involved a mock-design of a commercial civilian/military aircraft. The work involved grading weekly design reports, helping students with the design, and report-writing.