Has AI made certain roles at quant funds defunct?

It might be fair to say that AI will and has changed hiring more than any technology shift in our lifetime. A lot of roles we have previously hired for- trade monitoring, repetitive quantitative implementation, software plumbing around backtest and live execution code, etc are now being assigned to AI agents. I hear from others as well that hiring for roles such as Quantitative Trader and Quantitative Developer has dramatically shrunk, and these roles are likely to disappear in the near future. One hears about PMs at the big multistrats having to choose between more tokens and more engineers, and unequivocally going with more tokens. At the same time, AI engineers who can build AI-native infrastructure (the harness around the models) are becoming more valuable, as are those with strong financial intuition who can contribute to broader parts of the strategy development pipeline.

How is AI changing hiring for quants- both entry-level and seasoned ones?

Having spoken to lots of funds, the hiring for entry-level quants has also gone down. Some funds I know have indeed shut down their quant internship programs that used to have students work on self-contained quant projects for a couple of months. A lot of projects that I assigned to entry-level and junior quantitative researchers that were done over weeks, are today completed in minutes, and much more thoroughly and correctly, by AI agents.

At the same time, the need for quants with good judgment of what works and what doesn't, and solid modelling principles, is increasing. Everyone knows that if you just ask an LLM to generate alpha, it will come up with the most consensus alpha signals that have either decayed or are heavily crowded. The way that quant funds are using these tools today is feeding in their own priors into these models, and very strict criteria for what the output should look like. To this end, they are hiring mostly seasoned and experienced quants and PMs.

How have these recent trends changed hiring at FMI?

At FMI, we still uphold the modelling principles that we started off with- the difference is that today, they are followed by both agents and humans. Unlike most other funds today, we continue to hire and work with PhDs with little or no finance experience. We saw that hedge fund crowding was getting worse even before the rise of LLMs and is getting even worse with their rise. Working with radically contrarian, first-principles thinkers who bring insights from other disciplines is our way of escaping crowded trades and LLM-consensus ideas. If raw analytical horsepower is becoming commoditized, what becomes scarce is originality, judgment and persistence. Jim Simons famously said that he rates doggedness over intelligence as a quant skill, and I think that has never been more true.

How might these trends change the structure of a quant fund or pod?

The traditional quant fund model is of a CIO/PM supported by quants or analysts. The latter do the grunt work and number crunching, but have limited responsibility or skin in the game. That model might disappear and we might have PMs and associate PMs only, supported by an army of AI agents. Now since many PMs came through the ranks and started off as quants or analysts, the question remains of what the breeding ground for the next generation of PMs will be. I don't think anyone knows the answer to that.

Similarly, AI can run the risk scenarios; it cannot sit on the Investment Committee, vote on whether the firm should take that risk, and be accountable for the consequences like the CRO does. The compliance function (CCO) too is required to be a human, by law. In short, any role with accountability and liability is here to stay- that includes PMs, CIOs, CROs, CCOs and others of that nature.

Do these trends help or hurt startup funds?

The emergence of LLMs has injected some life to the quant fund startup ecosystem, and should help make the quant landscape much more competitive. LPs would often ask us how with a sub-10 person team we compete with the Citadels or Two Sigmas of the world, with their hundreds of PhDs and the massive edge in terms of the data and compute resources. They still retain the latter edge, but the dramatic reduction in the cost of intelligence means that smaller funds like FMI can compete much better than before.

People in Silicon Valley have been speaking of the potential emergence of a one-person unicorn for the last couple of years. I truly believe we are in an era where it might not be impossible to have maybe not one but a 3-4 person fund running billions of dollars, commanding an army of agents underneath them. Indeed, speaking to a couple of funds who are just starting out, they are already starting off with very lean, AI-native company structures with agents doing a majority of the leg work.