Shafaq Bajwa’s research and professional background bridge technology, data science, and inclusive education. Her work centers on developing accessible digital tools, such as interactive games and learning platforms for neurodiverse students. With a foundation in software engineering and data analytics, she integrates technical precision with pedagogical insight to address learning challenges. Her projects reflect a passion for leveraging data-driven design to support educational equity.
In the world of software development, data models are built on logic, efficiency, and predictability. Yet, the human users these systems serve are often anything but. This gap between clean code and complex human experience is driving a new conversation in technology, one that questions the limitations of purely quantitative analysis.
For data scientist Shafaq Bajwa, this conversation prompted a career move from developing banking applications to supporting students with special educational needs at Moorcroft School in London.
With an MSc in Data Science and a background in software engineering, Bajwa’s experience includes building systems where user adoption is critical. Her transition into a special needs classroom provides a unique lens on the unseen data that shapes human interaction—the subtle, contextual cues that algorithms often miss. This shift highlights a growing need to embed deeper, more empathetic understanding into the technology that shapes our lives.
Developing applications for large-scale use, such as paperless banking, often requires simplifying user behavior into predictable pathways. However, this approach frequently overlooks the nuanced realities of human interaction with technology. These simplifications can ignore critical factors that determine a product’s success or failure in the real world.
Bajwa notes that during her time in corporate tech, data models tended to disregard user hesitation, trust, and digital confidence. “Models assumed that all users would follow logical steps and adapt quickly, but in reality, many struggled with fear of error, language barriers, or resistance to change,” she explains. This observation aligns with industry findings that companies with a strong focus on design and user experience see tangible business results, including reports of 32% higher revenue growth.
These elements are difficult to capture in standard datasets but are crucial for effective technology design. “These emotional and behavioral nuances rarely fit neatly into datasets, yet they often determine whether a system succeeds or fails in real use,” Bajwa adds. This underscores a broader industry recognition that human-centric approaches are essential, as research shows 87% of employees believe empathy is a key driver for better outcomes and innovation.
For a traditional software engineer, the world is defined by discrete data types like integers and strings. However, in a special needs classroom, the most critical information is dynamic and relational. Translating these observations into a technical framework requires a new vocabulary that moves beyond binary success or failure states.
If creating a data dictionary for her work at Moorcroft School, Bajwa says she would move beyond simple variables. “I might include terms that reflect the humanity of learning behaviors: not binary variables, but shifting states of attention, emotion, and curiosity that can vary minute by minute,” she states. This approach is reflected in emerging frameworks like the Lived Experience Centered AI Framework (LEAF), which aims to integrate human context into AI development.
This new dictionary would redefine terms like ‘error’ as data to be interpreted with empathy, not just a state to be fixed. Bajwa emphasizes, “These are not discrete data types; they are dynamic, relational states that serve as a reminder that learning systems must interpret context, not just input.” This parallels advancements in fields like digital psychiatry, which use AI to interpret non-linear patient data through passive monitoring, uncovering patterns outside of conventional reporting.
The intersection of structured programming languages like Python and SQL with the unpredictable, often non-verbal communication of a special needs classroom presents a significant translation challenge. The goal is to build solutions that respond to human needs without stripping them of their essential emotional context. This requires treating subtle cues as valid data points.
For Bajwa, this process involves careful observation and interpretation. “In a special needs classroom, where communication may not always be verbal or predictable, I look for patterns of eye contact, gestures, engagement levels, and treat them as data, but without robbing them of the emotional context that gives them meaning,” she says. This method of gathering qualitative insights is gaining traction in AI development through techniques like ethnography-driven prompt engineering.
The ultimate aim is to create technology that is responsive to the individual. According to Bajwa, “The translation here is the empathy to build technology that first listens to the learner and second codes the response.” This human-first approach is similar to how researchers use large language models in a “conversational” process to analyze qualitative data, moving from basic summaries to deeper thematic understanding.
Traditional project management, with its emphasis on fixed scopes, timelines, and deliverables, often clashes with fields where progress is non-linear and deeply personal. In education, particularly with individual students, success cannot always be mapped onto a Gantt chart. This reality challenges the core principles of conventional management frameworks.
Working with students has reshaped Bajwa’s perspective on success metrics. She notes, “Success is measured by meeting deliverables on schedule in traditional project management versus in education, where it is measured by each learner at their unique pace.” This shift in thinking resonates with comprehensive performance measurement tools like the Balanced Scorecard (BSC) framework, which evaluates progress across multiple dimensions beyond the purely financial.
This experience has highlighted the limitations of rigid planning in human-centered work. Bajwa concludes, “The most impactful outcomes often arise outside the original plan, and human development does not follow a Gantt chart.” This reflects a broader challenge in business when it comes to translating qualitative outcomes into quantitative metrics, requiring a more flexible definition of value.
While a data scientist’s toolkit typically includes databases like Oracle and query languages like SQL, an environment centered on human learning demands a different set of instruments. In the classroom, the most valuable data is often qualitative, gathered through direct interaction and observation rather than automated dashboards.
Bajwa’s primary methods for data gathering have shifted significantly. “In the classroom, my essential tools have become observation, empathy, and reflection rather than databases or dashboards,” she explains. This hands-on approach is critical in fields where nuance is paramount, such as mental healthcare, which has a deep reliance on the clinician-patient relationship that AI struggles to replicate.
The information collected through these methods provides a depth that quantitative metrics often miss. “These tools aren’t digital, but they’re powerful in capturing the human data that truly defines learning progress,” Bajwa says. Her focus aligns with research that is grounded in the lived experiences of individuals to better understand the psychological impacts and risks of technology.
In software development, a ‘bug’ is an error—a deviation from the intended function that must be identified and fixed. However, in a learning environment, what might look like an error is often a critical data point. A student’s struggle can signal a moment of growth, connection, or a need for a different approach.
This context has completely altered Bajwa’s understanding of failure. “In software, a bug means something isn’t working as intended; in learning, a struggle often means growth is happening,” she states. This perspective is valuable for developing more resilient AI, as certain Sparse nonlinear modeling techniques are designed to create effective models from smaller or noisier datasets.
Instead of troubleshooting a person, the focus shifts to adapting the system. As Bajwa puts it, “Instead of debugging the learner, I focus on adjusting the environment or approach. In this context, what looks like failure is often just the first step toward understanding and connection.” This concept of systemic adjustment is mirrored in advanced AI ethics, where a Causal responsibility attribution framework evaluates both an AI’s decision and the design of its safety mechanisms when errors occur.
The daily observation of how students interact with learning tools offers a fundamental lesson for the broader technology industry. It reveals that human behavior is not static or uniform; it is shaped by attention, emotion, and environment. This variability is a reality that rigid technological systems often fail to accommodate.
According to Bajwa, the primary insight is clear. “The most important lesson from the ‘human code’ I observe daily is that every user is variable; no one interacts with technology in the same way every time,” she says. This principle is driving a push toward personalization in other fields, where AI is used to enhance treatment planning by forecasting an individual’s response to different interventions.
This understanding calls for a fundamental shift in design philosophy. “Technology should adapt to people’s changing needs, not expect people to adapt to rigid systems,” Bajwa emphasizes. Such a philosophy is crucial for creating more inclusive technology, inspiring frameworks designed to mitigate cultural bias in generative AI and produce more relevant outputs.
Learning to interpret the subtle, non-quantifiable data of students has permanently altered how a data scientist might approach any dataset. It fosters a practice of looking beyond the numbers to see the human stories they represent. This perspective encourages critical questions about what is missing, who is represented, and what the real-world impact is.
Bajwa’s approach to data analysis is now fundamentally different. “I no longer see data points as isolated values but as traces of human stories, emotions, and contexts that numbers alone can’t capture,” she reflects. This shift from correlation to causation is at the heart of advanced analytical methods like Double Machine Learning (DML), which aims to uncover true causal effects in complex datasets.
This deeper lens applies to any analytical task, from finance to algorithm refinement. “Whether I’m analyzing financial data or refining an algorithm, I now look beyond accuracy and efficiency to consider empathy, bias, and impact because every dataset, at its core, reflects human behavior, not just information,” Bajwa states. It is a perspective that aligns with emerging AI systems designed to capture the narrative structure of lived experience, organizing data in a way that better reflects human memory and reasoning.
The insights gained from environments that prioritize human complexity offer a valuable directive for the future of technology. As data science continues to evolve, integrating the ‘human code’—with all its nuance and unpredictability—is no longer just an ethical consideration but a practical necessity for building systems that truly serve their users.
Educational leaders warn that reliance on automated tools is eroding critical thinking and workforce readiness among the next generation.
At Global Business & Economics Journal, we contribute to the future of business and technology with news, innovative research, expert insights, and thought-provoking analysis.
