The Quiet Disciplines of High-Performance Labs: Talent, Modalities, and the Data-Native Future

1. Your work increasingly involves complex modalities, ADCs, oligonucleotides, peptides, and GLP-1 analogues. Do these require a different team-building approach than traditional small-molecule work?

They require a deeper bench.

In small-molecule work, a competent team can be built by hiring for HPLC, method validation, and impurity characterisation. The playbook is established, the regulatory pathway familiar, and most mid-career scientists can slot in and deliver.

Complex modalities break that model. An ADC is simultaneously a small molecule, a biologic, and a linker chemistry, and the critical quality attributes are the interactions between them, not their sum. Oligonucleotides require ion-pair chromatography and mass spectrometry at a level most labs have never done. Peptides sit uncomfortably between small molecules and biologics in regulatory expectations. GLP-1 analogues look straightforward until you start characterising degradation products.

The team-building implication is that you cannot build these capabilities by hiring alone; the talent pool is too thin. What works is building a core of scientists who hold the fundamentals, separation science, mass spectrometry, stability and growing them into modality specialists through structured exposure to actual problems, actual molecules, and actual regulatory submissions. Not training courses.

Companies winning in complex modalities accept that they must build capability internally; the market cannot give it ready-made. Those who try to acquire their way out pay a premium for people who leave in two years, and still have the same gap.

2. On talent, how do you think about hiring versus developing? In pharma, there is a real temptation to hire in expertise rather than grow it.

Both are necessary, but most organisations over-rotate on hiring.

Hiring solves a capability gap today. Developing solves it for the next five years. Hire only, and the team looks good on paper, but is fragile. Everyone is ready-made for their current role, and the institution has no muscle for growing people into the role they will need.

The half-life of a technique is shorter than the career of a scientist. The person you hired last year for LC-MS will lead biologics mass spec in three years, and something you cannot name today in seven. You cannot keep up by hiring alone. You have to build the development infrastructure.

That means technique deep-dives, cross-training between modalities, rotations between method development and validation, exposure to regulatory writing, the one almost nobody gets right, and actually giving people time to do these things. Learning does not happen in the gaps between deliverables. It has to be on the calendar.

3. Retention of top technical talent is harder than it has ever been. Beyond compensation, what actually keeps high-performing scientists engaged?

The work itself.

Compensation matters up to a point. Below market, you cannot hold people. At the market, you are in the game. Above market does not buy much once people are paid fairly, other things dominate.

What dominates is the work. High performers want problems that stretch them and equipment and systems that let them do real science. If a talented scientist spends sixty per cent of their time on admin tasks, data entry into systems that should have been replaced years ago, or method work beneath their capability, you will lose them not to a bidding war but to boredom. They may not tell you why.

4. Beyond the nature of the work itself, what factors influence the retention and engagement of technical talent in organisations, and how can leadership effectively address them?

The people, and the runway.

Technical talent stays where peers respect them, and leadership is technically credible. They leave where they feel talked down to, or suspect their manager does not understand what they do. “Leaders close to the work” matter as much for retention as performance.

The third is the runway, the visible path forward. High performers do not need a promotion every year. They need to see the organisation investing in capability, a next hard problem for them, and their development as part of the plan rather than a line item in next year's budget. Organisations that communicate “we see where you are going and are building toward it with you” retain people at rates compensation alone cannot explain.

5. You have led teams across the US, UK, and India. What changes when your team is multi-site and cross-cultural, and what stays the same?

Principles stay the same; communication protocols have to change completely.

In a co-located team, much coordination happens informally. People overhear things in the hallway. Problems surface over lunch. Context is shared without anyone writing it down. That invisible infrastructure does enormous work, and most leaders do not realise it until they try to operate without it.

In a multi-site team, none of that happens automatically. If you rely on informal patterns, sites end up working on different versions of the same project because, functionally, they are. Each develops its own interpretation of ambiguous instructions, its own sense of priorities, and its own view of “good enough.”

The move that works is over-investment in written clarity. Decisions, priorities, reasons behind decisions, and meeting outcomes are all written down. This feels bureaucratic to people who have only worked in single-site environments. It is not, it is the new informal.

The cultural piece is real, too. Cultures have different norms around disagreement and bad news. Someone who would push back immediately in one context may stay silent in another. Cross-cultural leaders must create explicit invitations for the quiet voices and explicit permission for disagreement; otherwise, you optimise for the loudest cultural defaults in the room.

6. AI and digitalisation are reshaping pharma analytical work rapidly. How does that change what you look for when building a team today versus five years ago?

It changes the questions, not the criteria.

Five years ago, interviewing an analytical scientist, I looked for method development depth, separation science fundamentals, and regulatory fluency. I still do. The new question is: what is your relationship with data?

By that I mean: do you understand what your instruments are actually generating beyond the peak report? Can you work with raw data, not just summary files? Are you comfortable with basic scripting? Do you consider data integrity at the audit-trail level, not just the final chromatogram? These are not specialised skills anymore; they are basic scientific literacy.

AI is a different conversation. I am cautious about “AI-enabled laboratories” without disciplined data practices. AI does not rescue a team with poor data hygiene; it amplifies whatever is already there. If your data is inconsistently structured, AI will generate confident, wrong answers. If your scientists do not understand the limits of their data, AI-generated analyses will bypass the judgment protecting you.

The team-building implication is the opposite of the hype. Before adding AI tools, build a team that is data-native, one that treats data as a primary scientific output, handles it with rigor, and can reason about what a model is and is not showing them. Such teams benefit enormously from AI. Teams without that foundation are safer without it.

7. What is the biggest mistake technical leaders make when trying to build high-performance teams?

A. Mistaking good management for good leadership.

As technical leaders move up, the job broadens, including hiring, compensation, project portfolios, client escalations, and board reporting. These take time. What many leaders do without realising it is delegate not just the hands-on work but the technical engagement itself.

The result is a leader close to their people but distant from the work they are doing. One-on-ones go fine, talent reviews are thoughtful. But when a genuinely hard technical call lands on the leader's desk, they lack the depth to pressure-test it. They defer, rubber-stamp, or escalate something that should not need escalation.

The team notices this faster than the leader does. A team that learns its manager cannot engage technically stops bringing the hard problems upward. They solve what they can, route around what they cannot, and the leader manages a team whose real technical life is invisible.

The fix is not micromanagement, it is staying fluent. Read the raw data yourself once a quarter. Sit in on a troubleshooting session once a month. Hold your own in a technical exchange with anyone on your team. Not to second-guess them, to make real decisions.

8. Final question. Someone reading this is about to take over a technical team, a CDMO laboratory, a contract testing group, or an innovator biotech CMC function. What are the first three things they should do?

One: Spend your first month learning the data, not the org chart. Read actual reports. Sit in on actual reviews. Read the last six months of deviations and OOS investigations. You will learn more about where your team is good and where it is stuck than any number of intro meetings will tell you.

Two: Ask the team what they think is broken before you tell them. The people doing the work almost always know. What they usually do not know is whether the new leader wants to hear it. The first month is when you establish that.

Three: Protect something. Pick one thing that is under-resourced and obviously matters, a skill the team needs, a process eating their time, a capability gap and visibly invest in it. Early investments are signals. They tell the team what you actually value, more loudly than any kickoff speech.

Do those three things in the first ninety days, and you will have earned the credibility to do everything else. Skip them, and you will spend two years trying to build trust you could have had from the start.