Skip to main content
Key Contributions
- Served as the sole dedicated contributor to business development for new issuers, reporting to co-founder Asher Gottesman. Turned commercialization priorities into screening criteria, experiments, data requirements, and outreach plans; independently managed analysis, counterparty discussions, and recommendations.
- Organized core-team reviews with commercial input from Asher Gottesman and technical and financial guidance from Gene Itkis of MIT Lincoln Laboratory and Michael Lustig, former Senior Managing Director at BlackRock and Adjunct Professor of Finance at NYU Stern. Assessed applied cryptography, zero-knowledge applications, technical fit, liquidity, and risk.
- Designed a two-stage framework for real-world assets, stablecoins, DeFi, and infrastructure: test identity, liquidity, pricing validity, data completeness, and risk, then rank commercial fit and accessibility. Used issuer-level error analysis to refine qualification rules, exclude out-of-scope projects, and record evidence-backed reasons for each decision.
- Resolved issuer identities across CoinGecko, chain records, websites, white papers, and disclosures. Reconciled multichain listings and wrapped or bridged assets to remove duplicate opportunities. Built market and text features spanning liquidity, capitalization, activity, project age, chain affiliation, business models, and technical needs.
- Built a Python, Prefect, DuckDB, and Parquet pipeline for ingestion, standardization, entity mapping, feature preparation, and querying. Implemented scheduled refreshes, retries, persisted outputs, reruns, and historical backfills to support model training and repeatable shortlist updates.
- Trained a LightGBM pairwise ranking model on the core team's comparisons of eligible counterparties. Combined commercial, disclosure, and market features to prioritize approximately 800 potential partners. Repeated comparison, retraining, and ranking over two to four weeks, using uncertainty sampling to focus expert review on borderline candidates, near ties, and conflicting evidence.
- Applied issuer similarity analysis and Leiden community detection to discover overlooked projects with related business or technical characteristics, then brought qualified candidates into commercial review and ranking.
- Led Codex-assisted solution design and Python development. Used meta prompting and ChatGPT deep research to translate business objectives into analytical plans and development specifications; selected methods, debugged workflows, and verified issuer-level outputs and selection rationales.
- Screened 18,000+ token records into approximately 800 potential counterparties, 30 outreach priorities, and six proposed pilot counterparties. Discussed asset-verification needs, service arrangements, and technology licensing with prospects, and delivered the candidate list and commercial assessments to the board for technical and business evaluation.