Data Scientist — Blockchain Intelligence

We turn raw on-chain activity into trustworthy intelligence clustering addresses into real-world entities attributing them to services and actors and surfacing risk for compliance and investigations teams. Were looking for a data scientist who is as comfortable shipping a heuristic to production as they are designing it: someone who can move from a messy hypothesis to a working pipeline without waiting on someone else to wire up the data.

Youll work closely with our attribution and clustering leads on models and heuristics that run across billions of transactions and multiple chains (Bitcoin Ethereum Tron Solana and more).

• Design test and ship clustering and attribution heuristics and measure them with real precision/coverage metrics rather than vibes.

• Own your data end to end pull clean join and model large on-chain datasets without depending on a separate team for every query.

• Build and maintain the pipelines that take a heuristic from notebook to production including backfills incremental runs and validation.

• Investigate edge cases (mixers bridges exchange hot wallets consolidation patterns) and translate findings into repeatable logic.

• Partner with investigations and product to define what correct looks like and benchmark against ground truth.

• Prototype quickly then harden what works.

• 4 years building data science or data engineering systems that actually shipped (not just notebooks).

• Strong Python and SQL; comfortable with large datasets and the gotchas of joins dedup and skew at scale.

• Solid grasp of clustering graph/network analysis or entity resolution and a habit of validating results not just producing them.

• Ability to reason about precision vs. coverage trade-offs and defend your metrics.

• Self-directed: you can scope an ambiguous problem get the data yourself and drive it to a result.

You dont need to have used all of these but heres what youd be working with day to day:

• Databricks our lakehouse and processing backbone. Large-scale on-chain datasets are transformed and modeled here via Spark and SQL; most heuristics run as Databricks jobs against billions of transactions.

• Kafka real-time ingestion of on-chain and transaction data. New blocks and events stream in continuously so a lot of our work is designed to run incrementally rather than as one-off batch jobs.

• Python the primary language for everything from exploratory analysis to production heuristics and pipeline code.

• TigerGraph our graph database where addresses transactions and entities live as a network. Clustering traversals and relationship queries (who funds whom consolidation paths entity linkage) happen here.

Supporting cast youll likely touch:

• SQL everywhere for ad-hoc analysis validation and defining ground-truth datasets.

• Columnar / analytical stores (e.g. ClickHouse) for fast aggregate queries over large tables.

• Orchestration & scheduling for backfills and recurring pipeline runs.

• Git / GitHub for version control and code review we expect pipelines and heuristics to be reviewed like any other code.

• GCP as our cloud environment.

Small high-trust team. Youll have a lot of ownership and very little bureaucracy. We prototype fast measure honestly and ship.