Head Of Talent Acquisition At Thoughtworks Shares How To Make Recruitment A Data-Driven And Predictive Process With AI

Throughout my career, I’ve held senior leadership roles in global technology and consulting companies such as Thoughtworks, Cognizant, and Hewlett-Packard, where I led large, multi-country Talent Acquisition teams. Earlier in my career, I also served as a strategic Talent Business Partner, advising senior and executive leadership on workforce planning, organizational design, and people strategy.

Working at scale shaped my leadership journey — managing teams of over 50 recruiters, leading 600 to 2,000 hires annually, and consistently delivering record results in hiring efficiency, candidate experience, and organizational growth.

In addition to core Talent Acquisition leadership, I’ve built and led strategic programs in employer branding, diversity and inclusion, and campus recruiting, including initiatives that significantly increased the hiring of women and people with disabilities and expanded early-career pipelines across multiple countries.

Alongside my corporate roles, I also serve as an independent advisor to beecrowd, a global competitive programming and technical assessment platform used by universities and technology companies to evaluate software engineering talent at scale. In this role, I advise on talent evaluation frameworks, skills assessment, and the intersection between technical excellence and workforce strategy.

In recent years, my work has increasingly focused on how AI and data-driven recruiting are reshaping leadership, talent strategy, and the design of high-performing organizations in complex, multi-country environments.

Why AI Is Shifting More Than Just Tech

With AI, my role as a leader in Talent Acquisition evolved from primarily focusing on execution to architecting strategic talent systems that leverage AI to transform how organizations attract, assess, and engage talent. This shift is not just technological — it reshapes leadership responsibilities, organizational design, and the relationship between people and data.

Talent Acquisition is traditionally transactional, but this transformation makes it data-driven and predictive. Instead of relying on manual screening and historical averages, we now apply AI-based learning agents and predictive models that analyze large datasets to identify success patterns and recommend candidates with higher precision.

By learning from historical data, these systems handle the repetitive tasks so teams can get back to what they do best: building relationships and making strategic decisions.

Ultimately, leadership in an AI-first era is about designing decision architectures: creating systems where people, data, and AI collaborate to produce better outcomes. This means building ethical governance, clear metrics, and integrated feedback loops that allow organizations to anticipate needs, reduce bias in selection, and align workforce capability with long-term strategic goals.

How AI Is Transforming HR Workflows And Decision-Making

Whenever we overhaul a workflow with AI, we're guided by a clear principle: AI removes manual effort, structures quantitative signals, and surfaces insights, while humans remain responsible for qualitative judgment and final decisions.

Here are a few examples of overhauls we have done:

1. Survey analysis and organizational feedback loops

One of AI's most effective applications has been analyzing large volumes of survey data. Using multiple customized Gemini models, we process open-text survey responses to identify recurring themes, sentiment patterns, and emerging risks across teams and regions.

Rather than relying on manual tagging or anecdotal interpretation, AI allows us to parameterize feedback at scale and turn qualitative input into structured signals that inform where interventions are most needed. This has significantly improved the speed, consistency, and objectivity of organizational improvement planning.

2. Internal workflows and manual process removal

We have also applied AI to internal workflows to reduce manual, repetitive work. AI triages requests, structures unformatted inputs, and supports decision preparation across internal processes.

This has meaningfully reduced operational overhead, allowing teams to focus on higher-value work like problem-solving, stakeholder engagement, and continuous improvement instead of administrative execution.

3. Performance and evaluation processes

In internal evaluation and performance processes, AI plays a critical role in handling the quantitative side of analysis. AI searches, aggregates, and cross-references structured data points over time, ensuring evaluations rely on longitudinal evidence instead of recency or isolated events.

This allows leaders to focus on the qualitative aspects of evaluation — judgment, context, development conversations, and future potential — instead of data gathering. The result has been more disciplined, fair, and evidence-based assessments.

4. Knowledge sharing and alignment

We use NotebookLM as a shared knowledge layer to document processes, decisions, and recurring themes, ensuring fast, secure access to institutional knowledge. We transparently share information captured in this system where appropriate, which improves alignment, reduces misinterpretation, and creates a common factual baseline across teams.

5. Strategy, decision-making, and org design

At a strategic level, AI supports how we think about uncertainty, priorities, and trade-offs. Instead of prescribing decisions, AI helps surface patterns, quantify ambiguity, and test assumptions — particularly in areas like role design, organizational effectiveness, and change initiatives.

How Predictive Analytics Improves Hiring Decision Quality

In Talent Acquisition, our cycle time was already consistently strong, in the 30- to 35-day range, so for us, it wasn't about fixing speed. The real shift came when we decided to redesign the quality of the funnel rather than process velocity.

First, we trained the Talent Acquisition team in predictive analysis and skills-based workforce segmentation. Instead of optimizing for volume, we redesigned our intake and screening models to predict which profiles were most likely to succeed in specific roles, markets, and teams. This allowed us to intentionally reduce the number of candidates entering each stage of the process.

Practically speaking, we built predictive dashboards combining historical hiring outcomes, performance data, and market signals. Recruiters then used these models to focus only on high-probability skills, moving from quality through volume to quality through precision.

But again: AI should not replace human judgment, but sharpen it. We deliberately preserved high-touch human interaction in interviews, calibration discussions, and final decisions. This combination allowed us to deliver a better candidate experience, more thoughtful assessments, and fairer, more consistent decisions.

How Job Design Impacts AI-driven Hiring Outcomes

Before working with predictive models, I underestimated how much ambiguous, inconsistent, or inflated job descriptions could distort hiring outcomes and AI system learning behavior. With AI, job design and job description quality determine every downstream decision.

In fact, running predictive analyses on our historical hiring data uncovered a clear pattern. Variability in role definitions drove poorer outcomes — even more so than low candidate quality. In other words, roles with poorly specified skills, unrealistic requirements, or inconsistent seniority signals consistently produced lower conversion, higher late-stage rejection, and weaker early performance, regardless of the strength of the candidate pool.

This fundamentally changed our strategy. Instead of treating job descriptions as static inputs, we redesigned them as predictive artifacts. We used historical outcome data to distinguish requirements that correlated with success from those that only increased noise and ambiguity. Over time, this allowed us to simplify role definitions, remove non-predictive requirements, and focus hiring on a smaller set of high-signal skills.

Why AI Adoption Should Be Institutionalized, Not Piloted

The biggest disconnect I consistently see between AI’s promise and organizational reality is not technological, it is organizational and human. Most companies invest heavily in AI tools and expect improvements. But they keep the same incentives, the same hierarchical structures, and the same low data literacy across leadership.

The result is that AI becomes either underused or misused: powerful models producing insights that are not trusted, not understood, or not acted upon.

We addressed this disconnect by institutionalizing AI adoption rather than treating it as a series of pilots. We established a dedicated cross-functional working group focused on AI-enabled decision design in Talent Acquisition.

The working group operates with explicit, business-level targets. Its two primary objectives are to increase conversion rates across the hiring funnel by 50% through predictive segmentation and decision redesign, and to reduce interview hours by an additional 30% by eliminating low-probability candidates earlier in the process. These targets force the organization to focus not on experimentation, but on sustained, system-level performance gains.

Beyond metrics, the group is responsible for redesigning decision workflows before introducing automation. We map where critical hirings are made, who owns them, what data is required, and where human judgment must remain central. Only after the decision architecture is clear do we introduce AI to augment specific steps, rather than layering tools on broken processes.

We also embed governance by design. This includes human override, bias and drift monitoring, clear accountability for model outcomes, and ethical review of high-impact use cases. Without this, AI tends to scale both good and bad decisions simultaneously.

What It Means To Be AI-Ready In HR And Talent Acquisition

We treat AI less as a set of tools and more as a decision-making mindset. Being AI-ready in our organization means consistently asking better questions: what signals matter, where uncertainty exists, and how data can inform human judgment.

AI literacy is built through everyday work, not formal training. Teams learn by using AI to surface patterns, test assumptions, and structure decisions across surveys, evaluations, and operational workflows.

In practice, it shows up in how we make decisions: more evidence-based discussions, clearer assumptions, better documentation, and a shared language for working with uncertainty. AI doesn’t drive decisions, it shapes how we think before we make them.

Carla's Three-Layer Tool Stack

For AI-assisted execution, we use several categories of tools. For language understanding, job design, and structured knowledge support, we use Gemini, an internal AI Agent and NotebookLM, serving as a secure internal knowledge layer to document processes and provide fast, governed access to institutional knowledge.

For interview intelligence, we use BrightHire to capture structured interview signals and improve calibration and feedback quality.

Within our core recruiting platform, we use Greenhouse AI primarily for scorecard summarization and structured synthesis of interview feedback, always preserving full human control over selection decisions.

Why NotebookLM And ChatGPT Are A Must-Have Cognitive Infrastructure

I’m particularly fascinated with two tools today: NotebookLM and ChatGPT. I use them in very different contexts.

I use ChatGPT exclusively for personal purposes, outside my corporate environment. It has become my primary personal thinking and learning tool. I use it for everything from structured learning and writing to exploring ideas, planning, and reflecting.

The platform itself indicates I’m among its top users, reflecting how deeply it has become part of my personal cognitive workflow. Its main impact has been accelerating my learning, improving my clarity of thought, and helping me reason more rigorously about complex topics.

In my professional context, I’m most invested in NotebookLM. I use it extensively to document and structure my one-on-one meetings, track recurring themes, and maintain a consistent, evidence-based view of individual development and performance over time.

I share these notes and summaries transparently with the people I meet with, which makes this particularly powerful. It creates a single, aligned record of expectations, feedback, and commitments, and has significantly improved trust, fairness, and consistency in performance evaluation. It allows both sides to work from the same factual baseline, rather than from memory or subjective interpretation.

This has resulted in a much more disciplined and fair approach to performance management: better follow-through on development actions, fewer misunderstandings, and higher alignment between managers and team members.

What I value most about these tools is that they are not automation tools, but cognitive infrastructure. Used this way, they don’t replace judgment — they improve the quality, fairness, and transparency of how judgment is exercised.

How Talent Acquisition Will Change In The Next Five Years

Talent Acquisition will stop being measured primarily by speed or volume and will become a decision-quality function. Over the next five years, the role will evolve from executing hiring processes to designing and governing how organizations make talent decisions under uncertainty.

AI will increasingly handle scale, pattern recognition, and signal extraction — from job design to market matching — while Talent teams become far more precise and intentional in designing and executing selection processes. The competitive advantage will come from clearer role definition, better signal quality throughout the funnel, and more consistent, evidence-based selection decisions.

At an industry level, I believe we’ll see a clear split: organizations that treat AI as a mindset embedded in their operating model will fundamentally outperform those that treat it as a set of tools. The future of the function belongs to teams that design better decisions, not just faster processes.

How Leaders Can Adopt AI Intentionally In HR

For people in roles like mine, my advice is to stop treating AI as an optional add-on and start treating it as part of the modus operandi of how work gets done. At this point, resisting AI is a contradiction: the question is no longer whether to use it, but how intentionally and responsibly we embed it into everyday decisions and workflows.

At the same time, we must be explicit about what should remain human. AI is extremely effective at handling scale, structuring information, and reducing noise. When used well, it frees people to focus more — not less — on what matters: judgment, relationships, ethics, creativity, and meaningful dialogue.

In a broader sense, my advice to leaders is to see AI as a force that clarifies priorities. Successful organizations use AI to remove work that technology does better than humans, and deliberately reinvest that time and attention into higher-quality human interactions.

Instead of diminishing leadership, AI creates the conditions for more thoughtful, focused, and human leadership to emerge.

Follow Along

You can follow Carla's work in revolutionizing talent acquisition on LinkedIn.

More expert interviews to come on People Managing People!