Software Engineer II

• We will be conducting a comprehensive video rating study for both WhatsApp calling and WhatsApp media sharing. The goal of this study is to establish robust technical metrics that will enable iterative system improvements - helping with distilling videos and images that are representative of what people would share on WhatsApp.

Goal of the engagement:

• We want to create a corpus of WhatsApp images and videos, and their encoded versions, that can be used to
• Create technical metrics that we can use to understand the perceived quality we are delivering to our users
• Offline evaluation of algorithms to improve perceived quality, The main responsibility will be to generate this corpus of videos
• Evaluate various video/image datasets and get a set representative
• Generate a corpus of encoded videos from the reference set
• Develop a scalable pipeline to apply various post-processing algorithms on top of lab generated videos (Will require using FFMPEG, GPU shaders)
• Validate videos generated meet a certain bar
• Sanitize the ratings data coming from Appen raters (crowd sourcing) and do some sanity checks (we already have scripts to SUREAL normalization etc)
• A third responsibility is to compare the human ratings with current technical metrics and identify patterns where current technical metrics don’t capture human ratings. Generate more videos following this pattern so that we can improve technical metrics

Day-to-Day Responsibilities:

• Write Python scripts and implement large-scale automation workflows to generate videos across various quality levels.
• Interact with an internal “lab API” to build baseline video sets using real devices on the WhatsApp stack.
• Build automated workflows to introduce specific video quality artifacts/differences (e.g., exposure adjustments) using tools like FFmpeg.
• Perform manual spot checks on generated video samples to maintain data integrity and test quality.
• Ingest, review, and analyze video ratings and raw media returned from external rating vendors to generate reporting on technical metrics.
• Develop Python scripts to transform and format rating data for downstream machine learning application ingestion.

• Strong Python scripting skills with experience in large-scale automation and data formatting for machine learning pipelines.
• Experience interfacing with REST/internal APIs and working with media processing toolkits like FFmpeg to programmatically manipulate video properties (e.g., exposure, quality parameters).
• Analytical capability to interpret external vendor rating data, track quality metrics, and compile performance reports. Attention to detail for conducting manual quality assurance and spot-checking generated media.

Nice-to-have Skills:

• Background in video quality assessment metrics (VMAF, PSNR, SSIM) or audio/video streaming protocols.
• Experience working with real-device testing frameworks or mobile stack automation (WhatsApp stack specifically).
• Knowledge of data pipeline integrations for Machine Learning feature store inputs. Familiarity with AI coding tools to accelerate their work Former Meta ideal but not a requirement

Years of Experience:

• 3-5

Degrees/Certifications Required:

• Bachelor’s Degree required, * Candidates lacking Python proficiency, candidates who have only done high-level program management without hands-on scripting/automation capabilities, or those with no experience handling video/media formats and APIs.