Faran Ahmad

About Faran Ahmad

I am a Software Engineer with interest and demonstrated experience in building large scale AI/ML and Data Infrastructure / Platforms. I enjoy solving scalability, reliability and efficiency related problems providing critical contributions to the company\'s top line business metrics.Currently, I work within AI Infra PyTorch org within Meta on Inference Enablement for large scale generative and sparse recommender models onto heterogeneous hardwares.For 4.5+ years, I worked within the Ads ML and AI Infra - Feature & Training Data Infra Org at Meta where we built ML compiler driven Feature Engineering Platform and Infra that enables feature authoring using expressive pythonic language and does a multi-query optimizations to generate efficient streaming and batch data pipelines. It also applies privacy enforcement and serves these features (leveraging custom C++/Velox DAG engine) at a massive scale for ML inference/training within Ads ranking and delivery, Modern Recommendation Systems for FB/IG, Integrity, Commerce, Search, Shops etc. For a couple of years, I have also worked within Ads Realtime Delivery org where I contributed to building an innovative Search + SQL engine driven analytics platform optimized for providing estimates of very large search queries with strict SLAs (< 20ms). This also involved building large-scale custom stream processing and batch systems to store peta-bytes of data in online storage systems enabling us to achieve this. The platform Is used for multiple use cases within Ads targeting, Bidding, Pacing, Audience / Pre-Campaign / Delivery insights, Payments Risk etc.

Experience

1. Staff Software Engineer

   Meta

   Jun 2017 — Present

   AI Infra - PyTorch Inference Enablement for RecSysPost training model optimizations to scale sparse / sequential arch of the generative recommendation models for GPU distributed inference at massive scale (10+TB). Ads ML & AI Infra - Feature and Training Data Infra TL within Realtime Feature Infra team where I worked on improving feature freshness, enabling new feature paradigms and modernizing company wide feature infra stack to deliver huge product wins across Ads, Feeds, IG and Integrity teams- Rearchitected our realtime/streaming feature infrastructure for Ads Ranking to improve feature freshness from 10+ min -> 18%, Facebook global session ~2%, IG Session > 0.11%, Feed VPV >~2%)- Developed several key capabilities to modernize the feature infrastructure for Integrity teams at Meta, including support for new operators, feature sharing, and ensuring seamless integration with training platforms3. Ads Realtime Data Infra - Audience Infra- Dynamic re-sharding and Elias-Fano encoding of the data to deliver 30% storage optimization for 1+ PB data, 30%+ memory improvements, 20%+ CPU utilization and reducing the overall service start time from 1+ day to a few hours- Supported new search query patterns e.g. filtering and aggregating data based on fact tables that can be combined with search queries.

Education

• CSKM Public School, New Delhi

  High School, Science

• Indian Institute of Technology, Delhi

  Bachelor’s Degree, Computer Science

• Indian Institute of Technology, Delhi

  Bachelor's degree, Computer Science

• Air Force Bal Bharati Public School, New Delhi

  High School, Science