Skip to content

2025 - 04

Plan

  1. Strategy Monitoring: Finalize metrics, dashboards, and alerts; validate data and risk thresholds.
  2. Trading Strategies: Deploy arbitrage and market-making strategies with integrated balance risk management.
  3. ML Infrastructure: Enhance model versioning and prediction monitoring while enabling collaborative development.
  4. Recruitment: Onboard candidates for quantitative researcher role.

Progress

After two months of travel, the team took some time to rest and recover at the beginning of April. We put technical initiatives (items 1-3) on hold to focus entirely on recruitment, where we made substantial progress.

Recruitment Journey

The recruitment process has been truly enlightening. We've started to see the power of leveraging team members as organizational multipliers. Our funnel metrics were impressive:

  • 230 applications received via LinkedIn
  • 76 candidates completed our lengthy, highly technical assessment
  • Used AI to efficiently rank and select the top 10 candidates
  • Invited these 10 plus 5 particularly eager candidates for interviews

Our somewhat delayed response time cost us one top candidate who accepted another offer before our interview. The caliber of applicants was remarkable – some even received competing offers from firms like BlackRock.

One standout candidate had grown his personal capital from $50,000 to $2.28 million over three years through algorithmic trading, focusing primarily on Bitcoin, Ethereum, and Solana. He received competing offers from both a well-funded startup and Fidelity's risk team, yet maintained a humble and respectful demeanor throughout our process.

Trial Progress

We offered trial tasks to 11 interviewed candidates:

  • 3 either declined or didn't respond
  • 8 candidates were successfully onboarded for trial

Two candidates exited early:

  • The most senior candidate (48-year-old from Korea) quit after reviewing the task requirements
  • A younger candidate (26 indian male) who relied heavily on AI responses withdrew after struggling to deliver a basic proposal

The remaining six candidates are progressing at different rates:

1st Tier

  • Zi-tong: Chinese female, born 1997, Quant Researcher at crypto trading funds. Strongest domain expertise, direct experience, proficient with modern tools, demonstrates urgency (likely due to OPT extension needs).
  • Matt-Kutt: White male, born 1990, data engineer previously at Fractal and Lockheed Martin. Strong overall performer, executes quickly, excellent at understanding and clarification, marathon runner with notable persistence.

2nd Tier

  • Tian-xie: Chinese female, born 2000, recent University of Chicago graduate. Very respectful, faster than average, limited experience but demonstrates strong interest in learning.
  • Nate-Ander: White male, born 1994, current MIT online master's student. Strong scientific background but limited production development experience and modern tech stack knowledge. Slower than average execution.
  • Kayode: Nigerian male, born 1988, post-graduate researcher. Strong academic background but weak modern software engineering skills.

3rd Tier

  • Run-Paw: Indian male, born 2000. Extremely eager but relies heavily on AI responses without full understanding. Initially had serious communication challenges but has shown willingness to become more honest and direct. Demonstrates exceptional enthusiasm, but his learning velocity remains unclear.

Key Insights

Our multi-stage recruitment process has proven highly effective at identifying genuine talent. The AI-assisted screening of detailed assessments allowed us to efficiently process a large applicant pool, while the trial tasks revealed crucial aspects that interviews couldn't capture. Most notably, we found execution speed and quality to be strongly correlated and better predictors of success than credentials or background.

The competitive talent landscape requires us to optimize response times and offer packages, as evidenced by losing candidates to established firms. Our strongest performers consistently demonstrate a combination of domain expertise, technical proficiency, and authentic communication. Reliance on AI without genuine understanding proved a reliable negative indicator, while candidates who could implement solutions independently thrived regardless of their geographic or educational background.

Looking ahead, we should maintain our rigorous assessment while improving our interview evaluation criteria to better predict trial success. The trial phase validated our approach of prioritizing demonstrated skills over demographics or credentials alone. This insight reinforces that our focus should remain on finding candidates who combine practical execution ability with domain knowledge and communication skills – the true indicators of potential contribution to our team.