Kristin Gleitsman

Leadership in Biotech

Tag: complexity

Kristin’s Bookshelf: Working with Humans

By

On December 19, 2025

In Kristin's Bookshelf

These books didn’t emerge from a single chapter of my career, but they all circle the same truth: working with humans starts with upgrading yourself. Not through performance, but through honesty, emotional range, and perspective.

The Stone & Heen books (Thanks for the Feedback and Difficult Conversations) came into my life during a period where I was learning — repeatedly and sometimes painfully — that I can’t control all that much except how I relate to a situation. That shift from “I don’t like this, it shouldn’t be happening” to “what’s actually going on here, and what are my options?” was one of the hardest pivots I’ve ever made.

These books helped me step back from the visceral reaction and see more of the chessboard so I could choose my next move with intention, not just instinct.

A little later came the reckoning with power dynamics…

For most of my career, I’ve had a deep allergy to organizational politics — especially the kind where polish is valued more than substance. My whole professional identity has been built around getting sh*t done. First alone, then with a team, then through leaders I developed. Power, by contrast, felt performative, slippery, and vaguely repulsive.

That’s why 7 Rules of Power is a book I love to hate and hate to love, depending on the day. There’s an uncomfortable honesty in it: the kind you want to look away from but can’t. The rules aren’t aspirational, but they ring true. And learning to navigate the reality of how power works, without losing sight of the kind of culture I want to build, is probably a practice I’ll be refining for the rest of my career.

Which brings me to Unlocking Leadership Mindtraps…

This book reminds me how easily we flatten reality into false binaries — right/wrong, good/bad, smart/stupid — especially under stress. Humans crave simplicity, but leadership rarely grants it. We live in a messy, colorful world where multiple things can be true at once. The more I can hold that complexity, the more effective (and compassionate) I can be with others.

So while I put these books under the category of “Working with Humans,” they’re really about something deeper: how to work with yourself so you can work better with other messy, complicated, brilliant humans, most of whom you have very little control over.

These books helped me widen my lens, soften my reactivity, and engage the reality in front of me rather than the one I wish existed. And from that place, the work becomes more honest, more humane, and more impactful.

Illustration of a desk with a figure from a recent paper about Deep-learning-based gene perturbation

When Complexity Misleads

By

On August 9, 2025

In AI ∩ Bio

In this week’s AI ∩ Bio paper, we look at a multi-dataset benchmark where deep learning models trained to predict single-cell perturbation outcomes did not outperform two deliberately simple baselines: “assume nothing changes” and “predict the additive combination of single perturbations.” Performance improved when models were reduced to just their internal embeddings and a linear decoder—no deep layers, just a direct projection from embedding space to gene expression. In other words, the models had learned biologically meaningful representations during training, but using those representations to make predictions through a complex decoder often made things worse.

This result may not surprise those familiar with traditional machine learning approaches. As the bias–variance tradeoff reminds us: when signal is weak and noise is high, simpler models often generalize better and overfit less, leading them to outperform more complex alternatives.

This study revealed both conditions:

  • The signal from genetic perturbations is often modest or additive.
  • The noise in single-cell gene expression is high, due to dropout, amplification bias, and biological variability.

If your model must detect subtle shifts in expression, but your measurement system is saturated with noise, then even an optimal model will struggle to generate meaningful predictions. And a large model may do worse, because it fits spurious correlations that simpler models ignore.

AI ∩ Bio

Illustration of scientific paper with the title "AI ∩ Bio: Reading the Revolution"

Before adding more architecture, it’s worth asking:

  • Have we clearly defined our outcome?
  • Do we understand the system we are working with?
  • Can our data realistically support the question we’re asking?

Otherwise, we risk building a computational Rube Goldberg machine: impressive, fragile, and ultimately unclear.

Powered by WordPress & Theme by Anders Norén