Hi 👋 I'm Nazih Kalo.

Relevant Coursework: Advanced ML, Deep Learning, NLP, Big Data, Data Engineering

Awards: Facebook Hackathon 2019 – WebBuilder ChatBot - 1st Place Prize

1. Built all data pipelines, including indexing & decoding on/off-chain data from multiple chains using Airflow/Spark/dbt
2. Developed nft & wallet recommendation engines, leveraging wallet trading/minting history to power follow/content suggestions
3. Maintained all internal/external dashboards (incl. dune, internal), retention/growth insights, & analytics for partners on link3.to

1. Built & maintained data pipelines for the company's largest data extraction/scraping project, scraping 12M+ products from ~5000 ecommerce sites. Extracted data was parsed, categorized/normalized to fit into customers’ taxonomy.
2. Developed internal Payout Optimizer to dynamically adjust payout functions to hit target rates; reduced pay variance by ~50% and led to $90k savings/month
3. Deployed self-hosted data cataloging tool (Amundsen), improving data discovery across the company & significantly reducing analytics team onboarding time. Extracted & linked Snowflake, dbt, BigQuery, Tableau, & Salesforce metadata.
4. Reduced LiDAR labeling time 34% through 1) optimizing ML pre-labels in product, 2) developing a new labeling pipeline (isolating 2D/3D labeling stages). New 2D labeling pipeline reduced computer spec requirement & increased labor pool.

1. Product lead for company’s new ML based text-moderation product; scope included dataset management, model training/deployment, post-training optimization, and monitoring/maintenance of SLAs
2. Collaborated with the ML team to develop a human-assisted/in-the-loop model auditing system to identify model deficiencies and error patterns in production data. Improved model F-1 score by 24% with minimal additional training data.

1. Built data pipelines integrating internal & vendor data to reduce spend forecasts latency from 168 to 24hrs
2. Managed data for $50M budget for iPhone XR dev builds and identified $1M fraudulent invoices through my analysis.