Building and recruiting AI teams

From: aidotengineer

Heath Black, Managing Director of Product at SignalFire, discusses strategies for building and recruiting successful AI teams, drawing insights from SignalFire’s proprietary AI/ML platform, Beacon [00:00:17].

SignalFire’s Approach to VC [00:01:43]

SignalFire operates as the first VC built like a tech company [00:01:46]. Similar to how tech companies interview customers, SignalFire interviewed 500 founders to understand their pain points [00:01:50]. Based on these insights, they built AI/ML tools and portfolio success teams to address common challenges like going to market, recruiting, building leadership skills, and launching products [00:02:08].

SignalFire’s platform, Beacon, tracks over 650 million employees, 80 million companies, and 200 million open-source projects [00:02:33]. This data is used to build proprietary ranking systems and market insights, powering the firm’s speed and supporting invested companies [00:02:44]. The focus for recruiting involves filtering the right people, finding them in the right locations, nailing the right timing, and closing them with the right narrative [00:03:02].

1. Filters: De-credentialization in AI Hiring [00:03:20]

A significant trend in AI over the past decade is a stark “de-credentialization” [00:03:49]. AI startups are increasingly hiring engineers without PhDs or prestigious schooling [00:03:53].

• Decline in Traditional Credentials:

  • In 2015, 27% of engineer hires were from top schools, and 16% had PhDs [00:04:03].
  • By 2023, these numbers dropped to 15% and 7% respectively, representing about a 50% decline [00:04:09].
  • Even for research scientists, less than half (about 40%) have advanced degrees [00:04:29].

• Shift in Focus: This shift reflects a market change from focusing on foundational ML research to applying models to real-world usage, emphasizing ML Ops, product experience, and understanding user interaction [00:04:51].

• Talent Mobility: Historically, AI talent was centered at large tech companies like Google, Uber, Meta, and Apple [00:05:30]. Over time, this talent has concentrated into what SignalFire calls the “AI V-League,” which includes companies like OpenAI and Anthropic [00:05:41]. Large tech companies are now actively recruiting from these AI V-League companies [00:06:14].

  • Knowing where people are coming from and where they are going is crucial for filtering the right candidates [00:06:28]. For example, OpenAI has a positive flow of talent from DeepMind, while Cohere has a negative trend [00:06:39].

Takeaways for Filtering [00:07:01]

• Work experience now far surpasses education as the main aspect to consider [00:07:04]. Focus on a candidate’s body of work, including open-source contributions and projects built outside of class [00:07:17].
• Assess whether a PhD researcher is truly needed for a role, or if an experienced engineer would suffice [00:07:36].
• Consider removing or softening academic requirements from job postings to ensure the top of the funnel attracts candidates with relevant experience [00:07:51].

2. Location: Where AI Talent Concentrates [00:08:06]

Despite debates about San Francisco’s decline, data indicates it is not “dead” [00:08:12].

• San Francisco’s Continued Dominance:
  • San Francisco makes up about 29% of all startup engineers, slightly down from 33% in 2013 but ticking up since 2021 [00:08:23].
  • For big tech, 50% of engineers still reside in the San Francisco Bay Area [00:08:49].
• AI-Specific Concentrations: San Francisco leads the pack for AI talent, with 35% of all AI engineers residing there [00:09:00]. Seattle accounts for about 22%, and New York for 10% [00:09:11]. San Francisco alone has more AI engineers than Seattle and New York combined [00:09:16].
• Funding Concentration: These cities are “punching above their weight” in AI hiring [00:09:28]. San Francisco accounts for nearly 38% of all early-stage funding into AI startups, despite only having 26% of all early-stage funding in the U.S. total [00:09:48].

Takeaways for Location [00:10:13]

• Data, not social media, determines market status [00:10:17]. Location still matters even in a distributed world [00:10:22].
• San Francisco, Seattle, and New York are the premier locations for AI talent today [00:10:27].
• Monitor location and funding markets to identify where talent and capital are flowing [00:10:35].

3. Timing: Identifying Opportunities and Impact [00:10:43]

Finding the right person depends on talent and timing [00:10:46]. Timing involves finding people when they are most likely to leave and finding those who would up-level a team at a specific company stage [00:11:11].

• Retention Rates: Analyzing retention rates of AI V-League companies can help identify when individuals are likely to be “poachable” [00:11:32]. For example, Anthropic has a 66% 4-year retention rate, while Perplexity hovers around 43-44% [00:11:43].
• Generational Behavior: Different generations exhibit different job mobility patterns [00:12:16]. In 2023, nearly 27% of Gen Z left their jobs, more than double the rate of Gen X [00:12:25]. Within four years of graduating, Gen Z averages 2.2 jobs, compared to 1.1 for Gen X [00:12:39]. This can be partly attributed to Gen Z’s slower promotion rates, layoff market conditions, and a greater penchant for taking risks and betting on themselves [00:12:52].
• Historical Composition Tool: SignalFire’s “historical composition” tool provides snapshots of admired companies at different points in time [00:13:42]. It shows organizational structures, key hires (e.g., sales leaders who scaled from $1millionto$10 million, or the first three engineers who shipped a key product), and helps identify candidates’ risk profiles, motivations, and potential as “10x hires” [00:13:51].

Takeaways for Timing [00:14:26]

• Understand timing from both an outreach and impact standpoint [00:14:29].
• Know when competitors or admired companies are likely to lose people [00:14:34]. Track employee profiles for changes [00:14:41].
• Study job change patterns across generations or population segments [00:14:50].
• Identify when people join and leave companies to find potential “10xers” and candidates likely to join a company at your stage [00:14:59].

4. Narrative: Beyond Pay and Equity [00:15:07]

Heath Black refers to Kurt Vonnegut’s concept of story shapes to explain how a company’s narrative is crucial [00:15:11]. A company needs to understand its triumphs, current position, and future trajectory [00:16:13].

• Declining Reliance on Traditional Incentives: Historically, pay and equity were the primary components of a company’s narrative [00:16:23]. However, this is changing:

  • From November 2022 to November 2024, the average tech salary only increased by 1.6%, with a sharp decline in equity granted [00:16:35].
  • AI Engineers are a “hot ticket,” commanding a 5% salary premium and a 10-20% equity premium over other engineering roles, making them even more expensive [00:16:50].
  • Reliance on salary as a sole selling point is no longer viable [00:17:14].
  • Equity is also a less reliable draw; in Q2 2024, only 33% of people exercised vested shares, down from 55% a couple of years prior [00:17:33]. This is due to concerns about high valuations, the cost of liquid capital, and the rapidly shifting AI market [00:17:47].

• Essential Narrative Elements: Companies must offer more than just money and equity [00:18:11]. Key elements for a compelling narrative include:

  • A close-knit environment with opportunities to work with founders [00:18:15].
  • Collaborative teams and minimal friction for getting work done [00:18:20].
  • A big mission [00:18:25].
  • Opportunities for personal growth and career advancement [00:18:29].
  • Exploding markets [00:18:31].
  • Solving complex problems [00:18:34].

Takeaways for Narrative [00:18:45]

• Understand these non-monetary aspects of your company to avoid relying solely on salary and equity in your narrative [00:18:37].

Conclusion [00:18:48]

In a competitive landscape where many companies are “fishing in the same engineering pond,” recruiting data provides a crucial edge [00:18:51]. Just as data is used to build products and models, it should be used to build teams, as a team is a company’s most valuable product [00:18:59].

Key points for recruiting AI teams:

• Filter accordingly: Acknowledge the “de-credentialization” trend [00:19:13].
• Watch where people move: Location still matters [00:19:16].
• Identify the right time: Use data to determine optimal outreach times and understand employee tenure at companies [00:19:23].
• Craft a better narrative: Build a compelling story beyond just compensation [00:19:31].

For job seekers, understand where admired individuals go and how long they stay to gauge company treatment and future prospects [00:19:47]. Finally, know what you desire in your career arc [00:20:03].