Lavanya Sharan

AI Researcher

San Francisco, California, United States22 yrs 2 mos experience

Most Likely To SwitchHighly Stable

Key Highlights

PhD from MIT specializing in machine learning.
Inventor of a patented image processing pipeline.
Led significant machine learning projects at Netflix.

Stackforce AI infers this person is a Data Science expert in the Entertainment industry with a focus on machine learning and optimization.

Contact

Skills

Core Skills

Machine LearningData ScienceCausal InferenceData StrategyA/b Testing

Other Skills

OptimizationPythonSparkSQLMetaflowData VisualizationBayesian ModelsJupyter NotebookStatistical ConsultingStatistical AnalysisNLPMatlabResearchC++Computational NeuroscienceConference Speaking

About

I am a data scientist and a former academic. I am a member of Data Science & Engineering team at Netflix, where I apply machine learning techniques to help optimize workflows in the post production space. I received my PhD in Computer Science from MIT, where I specialized in human vision, computer vision, and machine learning. As an academic, I am most known for my work on understanding how humans and computers recognize materials (e.g., paper, glass) in real-world photographs. I have published in top-tier venues including Nature, and my work has been covered by media outlets including Fast Company and CNET News. After receiving my PhD from MIT, I completed my postdoctoral training at Disney Research, Pittsburgh. While at Disney, I applied my expertise as a vision scientist to optimize the end-user experience, for which I was awarded a US Patent and a Disney Inventor Award.

Experience

22 yrs 2 mos

Total Experience

4 yrs 6 mos

Average Tenure

10 yrs 2 mos

Current Experience

Netflix

3 roles

Staff Data Scientist, Content Data Science & Engineering

Promoted

Aug 2021 – Present · 4 yrs 9 mos

Data science and machine learning for the newly formed Content Member Value space
Note: Netflix introduced formal levels for individual contributors in July 2022. I had been operating in (unofficial) Staff capacity for several years prior across three business areas.

Machine LearningData ScienceCausal Inference

Senior Data Scientist, Creative Production Data Science & Engineering

Dec 2017 – Jul 2021 · 3 yrs 7 mos

Shaped localization strategy at Netflix through machine learning, causal inference, and optimization techniques. Wore many hats (data scientist, tech lead, product manager) to create quantifiable and significant business impact. Contributions organized by technical domain included:
[Machine Learning] Led end-to-end development and rollout of a machine learning (ML) product that is used to guide and scale content localization efforts across Netflix
Created a suite of machine learning models to predict per-language content demand, including development of novel features, target variables, and solutions for handling missing data. Written in Python and SparkSQL using Metaflow.
Developed methodology to quantify the business value of ML-informed content localization efforts, and demonstrated significant improvement over existing baseline
Coordinated cross-functional project across 5+ orgs: set technical and non-technical roadmaps, developed product rollout strategy, gave roadshow presentations, increased product adoption, and managed stakeholder relationships
Mentored and provided technical guidance to three junior and new team members
[Causal Inference] Supported product strategy decisions by extracting causal insights from observational data
Quantified return on investment (ROI) of go-to-market localization strategy using Bayesian structural time series models
Measured member impact of delays in dub language availability using bespoke statistical methodology
Investigated member impact of content availability in Kids profiles using cohort studies
[Analytics & Strategy] Enabled workflow efficiencies by developing data analyses, data visualizations, and overall data strategy
Formulated data strategy for measuring member impact of specific content themes and influenced the design of associated workflows across two orgs
Created Jupyter Notebook-based data visualization tool that sped up workflows for content maturity ratings by 2.5x

Machine LearningCausal InferenceOptimizationPythonSparkSQLMetaflow+1

Senior Data Scientist, Streaming Science & Algorithms

Jan 2016 – Nov 2017 · 1 yr 10 mos

Optimized streaming experience for Netflix members through A/B experimentation, statistical consulting, and analysis tooling development.
Designed and analyzed A/B experiments to optimize streaming quality across multiple business areas: adaptive algorithms, content encoding, and user interfaces
Influenced the development of VMAF, a perceptual video quality metric that won a 2020 Technical Emmy Award. Resulted in US Patent 20190295242A1.
Devised and served as primary contributor for a self-serve Python test analysis tool that accelerated the pace of testing by 3x
Communicated findings from A/B tests at forums across the company with a focus on data storytelling and actionable insights
Taught introductory statistical concepts to technical and non-technical partners via internal workshops and regular office hours

A/B TestingStatistical ConsultingPython

Insight data science

Fellow

Sep 2015 – Dec 2015 · 3 mos · Palo Alto, CA

Collaborated with URX, Inc. to build a keyword extraction algorithm for their mobile deep linking platform
Developed a keyword classifier for URX that achieved state-of-the-art performance on Crowd500 benchmark
Delivered Python code for classifier using machine learning (scikit-learn) and NLP libraries (nltk, gensim)
Recognized for project contributions on official blogs for both URX and Insight Data Science

PythonMachine LearningNLP

Massachusetts institute of technology (mit)

2 roles

Research Scientist, Dept of Brain & Cognitive Sciences

May 2014 – Dec 2015 · 1 yr 7 mos · Cambridge, MA

Co-authored 4 grant proposals, including one that resulted in a Google Research Award of $70,000
Awarded Outstanding Reviewer at 2015 IEEE Conference on Computer Vision & Pattern Recognition

Postdoctoral Associate, Dept of Brain & Cognitive Sciences

Oct 2012 – Apr 2014 · 1 yr 6 mos · Cambridge, MA

Conducted research on topics in human vision, with an emphasis on understanding mechanisms of peripheral vision.
Investigated what humans see in peripheral vision using experimental methodology similar to A/B testing
Generated visualizations of losses in peripheral vision using image processing algorithms in MATLAB
Received press in Fast Company, Visual.ly blog, and MIT News for peripheral visualizations of Boston subway maps

Carnegie mellon university

Instructor, Robotics Institute

Jan 2011 – May 2011 · 4 mos · Pittsburgh, PA

• Designed and co-taught a new graduate class on human vision with Prof. Alexei Efros

Disney research pittsburgh

Postdoctoral Associate

Sep 2009 – Sep 2012 · 3 yrs · Pittsburgh, PA

Conducted research on topics at the intersection of human vision, computer vision, and computer graphics, with an emphasis on applications
Invented and optimized via behavioral testing an image processing pipeline to generate 3D models of human appearance. Resulted in US Patent 13/407,606 and a Disney Inventor Award.
Demonstrated that motion blur effects in video games do not improve player experience, contrary to popular belief. Awarded Best Oral Presentation at 2013 ACM Conference on Motion in Games for this work.
Formulated guidelines for audio-video synchronization in television & film production based on usability testing. Recognized as a top paper at 2010 Conference on Applied Perception in Graphics & Visualization.

Massachusetts institute of technology (mit)

3 roles

Teaching Assistant, Dept. of Electrical Engg & Computer Science

Promoted

Feb 2009 – May 2009 · 3 mos · Cambridge, MA

• Taught laboratory sections for 6.02 Digital Communication Systems.

Teaching Assistant, Dept. of Electrical Engg & Computer Science

Sep 2004 – Dec 2004 · 3 mos · Cambridge, MA

• Taught tutorial and laboratory sections for 6.002 Electronic Circuits.

Research Assistant, Perceptual Science Group

Sep 2003 – Aug 2009 · 5 yrs 11 mos · Cambridge, MA

Conducted research on the novel topic of material recognition, and advanced knowledge in two fields: human vision and computer vision
Published in top-tier journals & conferences, leading to 2000+ citations: Nature, Intl. J. Computer Vision, IEEE CVPR
Received press in CNET News, Phys.org, MIT Homepage Spotlight
Co-authored 2 grant proposals that resulted NIH R01 and R21 awards of amount: $573,255
Conducted the first systematic study of how humans recognize material categories using behavioral testing
Built the first computer vision system to recognize material categories in images using MATLAB
Created the first benchmark image dataset for material recognition