Puspanjali Sarma: A Steward of Human-Centered Intelligence

Strengthening algorithmic fairness, explainability and long-term governance in AI platforms.

High-performing enterprises rely on one thing even more than speed: clear judgment. Modern systems record millions of signals every hour, yet leaders often feel overloaded. Metrics appear across screens, lines of SQL move through pipelines, and teams still wonder which decision can carry the business forward with certainty. This tension sits at the heart of the AI industry. It also sits at the heart of Puspanjali Sarma’s career.

Today, she leads as a Senior Manager – AI, Data Engineering & Analytics at ServiceNow, where she works with teams that create AI capabilities trusted by large global organizations. Her work supports ServiceNow’s mission to bring intelligence, clarity and reliable automation into enterprise operations. Her path toward this role began much earlier, in a home anchored in education, ethics and steady effort. These early experiences encouraged her to pursue work with purpose and to stand by principles even when the environment becomes demanding.

Let us learn more about her journey:

A Purpose Shaped by Observation

Early in her career, Puspanjali noticed a recurring pattern. Organizations had more data than ever, yet decisions often defaulted to hierarchy or instinct. Teams hesitated to trust their own systems. Insight existed, but judgment did not always follow.

“I wanted to connect raw data to meaningful judgment, without losing sight of the people behind the numbers,” she explains.

That motivation guided her toward AI engineering, analytics, and leadership roles where technology and accountability intersect. It also led her to write Strategic AI Leadership Through Data, where she explores how systems thinking, governance, and leadership must evolve together.

How She Thinks About Scale and Direction

Puspanjali plans with a long horizon and executes in short, deliberate cycles. She maintains an 18 to 24 month view of how AI and data capabilities need to evolve across data engineering, governance, MLOps, and user experience. At the same time, her teams commit tightly to the next 90 days. This balance allows progress without locking teams into assumptions that may not hold.

Her architectural philosophy follows the same logic. She favors modular, cloud-native, and vendor-neutral designs that can adapt as needs change. Across her career, this approach has supported the development of AI assistants, quality engineering agents, and synthetic data frameworks built on open, well-governed platforms rather than closed ecosystems. Synthetic data, in particular, enables teams to test and experiment safely without exposing sensitive information.

Tools evolve. Principles stay steady. She prioritizes trust, reliability, and business impact, and remains flexible about how those outcomes are achieved.

“I plan like a strategist but execute like a startup.”

Design Choices Grounded in Outcomes

Within her teams, architecture is never abstract. Design decisions are tied directly to business value and operational clarity. Assistants for analysts, data quality companions, and testing accelerators are built to reduce friction, not introduce new dependencies.

She pushes for plain language in technical discussions and consistently brings conversations back to outcomes leaders care about. Can this system be trusted? Will it scale responsibly? Does it make work meaningfully easier for users?

A Calculated Risk That Paid Off

One defining moment came when she proposed an AI-first approach to testing data pipelines. At the time, teams relied heavily on manual test case design. Release cycles slowed, and highly skilled engineers spent disproportionate time on repetitive validation work.

She advocated for a quality engineering companion that could interpret SQL, generate test scenarios, and validate pipelines on Snowflake. The goal was to improve confidence while reducing manual effort. Stakeholders raised concerns around audit readiness, accuracy, and governance.

Rather than push blindly, she led a controlled pilot with clear guardrails and measurable goals.
The outcome was tangible. Release cycles shortened significantly, defect leakage dropped in a meaningful way, and engineers were able to focus on higher-value design and analysis.

“Bold ideas work when they are paired with discipline and explained in business terms people trust,” she notes.

Turning Ideas into Enterprise-Ready Products

Puspanjali structures innovation through two clear modes. The first is idea time. Hierarchy steps back, and engineers, analysts, and product managers explore concepts freely. Many of the organization’s analytics and quality engineering innovations began here.

The second is implementation time. Governance, security, and legal partners engage early. Model documentation, access controls, audit logs, and human review become standard requirements, not afterthoughts.

The message is consistent. Creativity is encouraged. Accountability is non-negotiable. Both are required for enterprise AI.

From Project Delivery to System Thinking

As her responsibilities grew, so did her perspective. Earlier roles emphasized delivery, timelines, and scope. Over time, her questions changed. Which cost or revenue lines does this influence? Will users still value this in three years? How does it behave under pressure?

This shift led her to treat internal AI initiatives as products, complete with roadmaps, feedback loops, and long-term ownership. She works closely with leaders across sales, finance, marketing, and operations to translate AI into practical, everyday value. This approach aligns with ServiceNow’s broader focus on AI that strengthens real workflows rather than isolated experiments.

Staying Grounded Through Setbacks

Her path to senior leadership was not without friction. She has worked in environments where bias or misalignment challenged her values. Those experiences were difficult, but clarifying.

Her anchor remains consistent. “I want AI leadership to feel human, ethical, and accessible. Setbacks can slow progress, but they do not erase purpose.”

During challenging moments, she reflects on long-term impact: platforms built, teams developed, and systems that moved from concept to daily use. She also leans on mentors and peers who offer direct, honest perspective.

Mentorship as a Measure of Impact

Mentorship is not optional in her leadership model. “Mentoring is how I measure my own impact.” She supports students, early-career professionals, and mid-career women through candid conversations about opportunity, bias, and recovery from unhealthy environments. Inside her teams, empowerment is practical. Engineers and analysts present their own work to senior leaders. Skill mobility is encouraged. Data engineers explore ML. Analysts learn product thinking.

“Empowerment means giving real ownership and staying present when people stretch.”

Risk Awareness That Enables Creativity

With a background in risk management, she brings structured thinking to innovation. Her teams ask who could be affected unintentionally, how models behave when confidence is misplaced, and how tools might be used beyond their original intent.

Experimentation and production remain clearly separated. Sandboxes allow speed. Production systems demand monitoring, auditability, and clear ownership. This discipline reinforces trust rather than slowing progress.

Building Measurable Change

Across roles and organizations, her work consistently centers on quality, reliability, and productivity. She has led initiatives where AI generates and validates test cases across complex data ecosystems, improving consistency and reducing manual effort. She has guided the development of assistants that allow business users to explore data safely through governed queries.

Evidence matters. Faster cycles, lower defect rates, and better use of expert time guide decisions. Systems that fall short are improved or retired.

The Future of Women in Enterprise AI

She sees women playing an increasingly central role in AI leadership, particularly in areas like governance, ethics, and long-term risk. These conversations shape reputation and strategy.
More women will lead AI centers, serve as Chief Data and AI Officers, and found advanced technology companies. Leadership styles that emphasize clarity, collaboration, and steady judgment will become the norm.

The milestone she looks forward to is simple. A woman leading a major AI initiative will feel unremarkable.

Advice for Women Building Their Path

Her guidance is direct.

“Keep your ambition alive. It is a strength.”

She urges women to pay attention to culture. Persistent dismissal, normalized bias, or declining wellbeing are signals to take seriously. Document impact, build alliances, and leave environments that do not respect values. Prestige is never worth personal cost.

Careers unfold in seasons. Leadership is built over time, through consistent choices.

The Qualities That Will Matter Most

Future leaders will be fluent across AI, business, and people. They will ask better questions, create spaces where ideas compete fairly, and turn complexity into direction.

She highlights three qualities. Courage. Curiosity. Clarity.

Above all, they will lift the women who follow.

A Leader Focused on Trust

Across every stage of her journey, one thread holds. Puspanjali brings discipline, compassion, and practicality to AI leadership. Her work strengthens how enterprises decide and act. Her teams build systems that support real people. Her leadership invests in inclusion and mentorship.

“AI works best when it helps people think with clarity and confidence,” she says. “My role is to build teams and systems that make that possible.”