Jacques Bernier
Staff Software Engineer | Search, Recommendation And DevOps
San Francisco,CA
Summary
Staff Engineer with expertise in building robust and reliable search and recommendation systems powered by machine learning at scale. Able to build production ready machine learning solutions from the ground up.
Support teams to improve operation, incident response and engineering reliability. Experienced in building engineering roadmaps and collaborating with cross-functional teams to plan and execute technical ML projects, ensuring alignment with business goals and strategic objectives.
Excellent communicator and force multiplier with ability to teach, mentor and quickly resolve issues.
Overview
10
10
years of professional experience
Work History
Staff Software Engineer
Objective Inc
San Francisco, CA
06.2023 - Current
- Designed and build self-service ingestion and control plane APIs to allow customers to onboard and send their data without any interaction from engineering team.
- Redesigned search API for cost efficient by using KServe to deploy embedding and ranking models using GPUs while deploying API server to cheap hardware. Reduced cloud infrastructure cost by 60%
- Rebuilt batch embedding and indexing pipeline to be realtime using Temporal Workflows, Kserve and Qdrant to reduce data latency from 8 hours to 1 minute on p99.
Senior Software Development Engineer | Tech Lead
Twitch (Amazon)
San Francisco, CA
08.2018 - 10.2022
- Led team of software engineers and applied scientists to build state of the art search and content retrieval system for all of Twitch content.
- Saw through multi-year migration from 3rd party search vendor to first party system leading to 2.2x lift in conversion.
- Designed a 2 phase search system at Twitch to support billions of documents and more than hundreds 100k requests per second.
- Worked along side applied scientists to build ML models for ranking, query understanding, query suggestion and vector search using embeddings.
- Designed framework agnostic (tensorflow, pytorch, xgboost) training pipeline for search ranking, query understanding and query suggestions using AWS Step Functions, AWS Batch, AWS Glue (pyspark) and Lambda
- Designed online search re-ranking system built on top of AWS Sagemaker (bring your own model) and DynamoDB as feature store.
- Optimized feature processing in python using profilers to reduce latency and training time.
- Implemented configuration based request flow to support experimenting with multiple re-ranking and retrieval models.
- Implemented affinity and similarity content retrieval system using embeddings and AWS OpenSearch scaling to billions of documents.
- Built monitoring using business and conversion metrics to prevent ML model regression.
- Wrote operational readiness guide to help feature teams across Twitch successfully launch machine learning based products and features to production.
- Mentored engineers, taught skills in python, go, AWS, and system designs and worked to improve overall team performance.
Senior Software Engineer
Pandora
Oakland, CA
12.2015 - 08.2018
- Prototyped and led innovative machine learning project to group and disambiguate music tracks and artists at scale which helped recoup millions of dollars in minimum guaranteed revenue and enable interactive features in product for 96% of playable catalog.
- Productionized pairwise ML model that used multiple audio fingerprinting algorithms along with text metadata to disambiguate entire music catalog containing 100M tracks.
- Built labeling tool to allow curators and contractors to provide training data for ML model.
- Deployed risky data migration to move from 100% manually curated music catalog to fully automated catalog with identification, de-duplication and disambiguation mechanisms powered by ML model.
- Trained and mentored engineers
Software Engineer
Rdio Inc.
San Francisco, CA
07.2014 - 12.2015
- Refactored and maintained legacy music metadata and rights ingestion system to support 70M tracks and rights in 85 countries across >250 music providers.
- Migrated payment provider to support growing business and international expansion.
Education
Bachelor of Science - Software Engineering
Laval University
Quebec City, Canada 08.2010 - 05.2013
Skills
Machine learning systems
Communication and presentation
Elasticsearch, Redis, Qdrant
Kinesis, Kafka
AWS Infrastructure & networking
Python, Go
DynamoDB, Redshift, CoackroachDB
Timeline
Staff Software Engineer
Objective Inc
06.2023 - Current
Senior Software Development Engineer | Tech Lead
Twitch (Amazon)
08.2018 - 10.2022
Senior Software Engineer
Pandora
12.2015 - 08.2018
Software Engineer
Rdio Inc.
07.2014 - 12.2015
Bachelor of Science - Software Engineering
Laval University
08.2010 - 05.2013