Sina Vafadar

Curriculum vitae

Building data infrastructure, analytical systems, and research frameworks across blockchain protocols and incentive mechanisms. Based in Canada.

snvfdr@gmail.com · linkedin.com/in/SinaVafadar · Canada

Experience

2025 - Now

Blockchain Data Engineer

DoubleZero Foundation · Remote

2024 - 2025

Blockchain Data Engineer

LayerZero Labs · Vancouver, BC (on-site)

  • Quantitative model: Designed and implemented an incentive model to reward DVNs for verification.
  • Revenue model: Documented LayerZero’s revenue model, ran analysis, and proposed key improvements.
  • Data pipelines: Built 5+ pipelines to share data internally and with third parties like DefiLlama and Dune.
  • API endpoints: Developed 8+ API endpoints powering LayerZeroScan and internal tools.
  • Infrastructure: Migrated pipelines to Airflow and resolved 15+ critical data inconsistencies.
  • Other: Built 20+ dashboards and onboarded a data analyst.
stack

LayerZero stack

third party pipeline read surface

The data stack at LayerZero Labs On-chain → DynamoDB; CoinGecko → DynamoDB; DynamoDB → Overdrive; Overdrive → Postgres; Overdrive → Snowflake; Postgres → Spacetime; Spacetime → Frontend; Postgres ↔ Maestro; Snowflake ↔ Maestro; Snowflake → Hex; Snowflake → AWS S3; Snowflake → Dune; AWS S3 → Dune; AWS S3 → DefiLlama. On-chain CoinGecko CoinMarketCap DynamoDB Overdrive AWS ECS Postgres AWS RDS Snowflake Airflow Hex AWS S3 Spacetime AWS ECS Maestro AWS ECS Dune DefiLlama Frontend
On-chain events and price feeds land in DynamoDB. Overdrive loads Postgres and Snowflake, where Airflow schedules the rest. Spacetime serves the frontend; Maestro reads and writes both stores; S3 is the handoff to Dune and DefiLlama. Drag the graph sideways to see it all.
2023 - 2024

Blockchain Data Scientist

OpenBlock Labs · San Francisco, CA (remote)

  • Quantitative model: Co-developed an incentive model on Sui (TVL increase from $343M to $1.3B).
  • Elasticity model: Built a model to analyze the link between incentives and TVL growth.
  • Wash trading: Created a model to detect wash trading on DEXes, flagging over $20M in suspicious activity.
  • Toxic flow: Built a model to assess toxic flow on DEXes and its impact on LPs.
  • Sybil detection: Implemented a model that uncovered 500K+ Sybil addresses used for incentive farming.
  • Looping detection: Designed a model to detect $100M+ in looped borrowing in lending protocols.
  • Project management: Owned the Sui project and NAVI incentive program, headed the lending protocol team.
  • Other: Onboarded and trained 2 data scientists, and built 10+ dashboards.
stack

OpenBlock stack

third party pipeline read surface

The data stack at OpenBlock Labs Sentio → AWS S3; AWS S3 → Frontend; QuickSight → Frontend; Sentio → Deepnote; DefiLlama → Deepnote; Dune → Deepnote; Snowflake → Deepnote; AWS S3 ↔ AWS Glue; AWS Glue → AWS Athena; AWS Athena → QuickSight; AWS S3 ↔ Deepnote; AWS Athena ↔ Deepnote; QuickSight → Deepnote. Frontend AWS S3 Sentio DefiLlama AWS Glue Deepnote Dune AWS Athena QuickSight Snowflake
Sentio indexes into S3, which serves the frontend and is crawled by Glue for Athena and QuickSight. Deepnote is the bench: it reads S3 and Athena and pulls in Dune, DefiLlama, Sentio and the partner Snowflake. Drag the graph sideways to see it all.
2023

Data Scientist Intern

Mastercard · Vancouver, BC (remote)

  • Collaborated with Mastercard AI Garage to develop a leading indicator for detecting fraud in cryptocurrency on Mastercard transaction data. Subject to NDA, no further details.

Education

2021 - 2023

M.Sc. Computer Engineering

University of Alberta · gpa 4.0 / 4.0

  • Thesis: Condorcet Attack Against Fair Transaction Ordering.

Publications

Vafadar, S., & Khabbazian, M. (2023). Condorcet Attack Against Fair Transaction Ordering. 5th Conference on Advances in Financial Technologies (AFT 2023).

doi:10.4230/LIPIcs.AFT.2023.15