From Factory Floor to AI Frontier: Harshit Kikani’s Journey at Amazon

Harshit Kikani has traversed a professional path from mechanical engineering on the manufacturing floor to leading artificial intelligence initiatives at Amazon. With an MBA and dual master’s degrees in mechanical engineering, as well as extensive roles in engineering and management, Kikani’s expertise bridges operational grit with digital innovation.

Currently serving as a senior manager for product management at Amazon, his multi-faceted experience is now central to workforce development efforts in the era of intelligent automation. Kikani’s transition embodies a larger trend reshaping global industries: the transformative merger of operational know-how with advanced AI technologies.

This shift is driving new standards in workforce training and operational excellence, as organizations look to digitize legacy processes without leaving behind the human expertise that underpins their success. In Amazon’s AI innovation environment, Kikani’s journey exemplifies how the lessons of the shop floor are influencing the shape of tomorrow’s digital workplace.

Manufacturing roots shaping AI innovation

Kikani’s early years managing assembly lines and troubleshooting factory operations have defined his approach to AI at Amazon. “When I started as a mechanical engineer working on turbocharger and exhaust system assembly lines, I learned that the best solutions come from deeply understanding the problem you’re trying to solve,” he recalls. He stresses that technology, whether robotics or language models, must serve the people who use it, not the other way around.

This perspective has led to a problem-first mindset in AI innovation, where solutions are grounded in the reality of day-to-day work. “The speech and text recognition tools we’ve integrated aren’t just impressive technical achievements; they’re designed to enhance communication and accessibility for real people doing real work.” Such views align with industry recognition of the role of virtual reality and digital technologies for validating workflows and improving worker training across complex environments.

Kikani’s focus on embedding operational insights into technical development reflects an ongoing shift—one that privileges domain expertise and collaborative design alongside algorithmic prowess.

Lessons in operational excellence

The pursuit of reliability, measurement, and improvement learned in manufacturing continues to inform Kikani’s AI leadership. “Reliability at scale” is critical, he notes: “In automotive manufacturing, a single point of failure can shut down an entire production line.” This same vigilance is applied to machine learning solutions, where contingency planning and robust monitoring are paramount.

Reflecting on operational cultures, Kikani describes, “Continuous improvement translates to continuous model monitoring, A/B testing, and iterative refinement within AI systems. The cost savings I helped drive through analytics at Amazon came from this same relentless focus on measurement and improvement.”

The need for collaboration between varied experts is also foregrounded, as AI solutions depend on the cross-functional integration of data scientists, business users, and technical staff—a view supported by the emphasis on strategic team structures in modern enterprise environments.

Augmentation, not automation

Kikani advocates for intelligent augmentation, rather than mere automation, as the guiding principle for integrating AI into business operations. Recalling hands-on experience with robotics, he argues, “A skilled machinist could detect subtle vibrations or sounds that indicated tool wear. A robotic cell that’s too complex to troubleshoot becomes a liability, not an asset.”

This philosophy is reflected in his approach to AI-powered training and operational tools. “Large language models and machine learning algorithms shouldn’t replace human expertise; they should scale it, democratize it, and make it accessible where it’s needed most.”

Explainability and user-centric design are prioritized so that business leaders need to understand not just what the AI recommends, but why. Such priorities mirror broader advancements in the field, such as scalable and interpretable LLM solutions that enable effective decision-making.

Solving workforce training challenges

One of Kikani’s pivotal projects combined operational analysis and AI-driven solutions to address a workforce training bottleneck. “We had experienced trainers who were excellent at their jobs, but they couldn’t scale. Quality and safety were suffering, and onboarding times were extending beyond acceptable limits,” he explains. Through process mapping and standardization, Kikani’s team reimagined training bottlenecks using artificial intelligence.

“We developed an integrated training platform using large language models that could deliver personalized instruction, answer common questions, and adapt to individual learning speeds.” The key insight was to free [trainers] from routine information delivery so they could focus on high-value activities.

This approach also incorporated speech and text recognition tools to serve diverse learning needs, reflecting ongoing research into the impact of virtual reality platforms for high-stakes, hands-on practice across industries.

Preparing workers for automation

Kikani’s experience bridging plant teams and AI teams reveals similarities and differences in preparing workforces for automation. “Both manufacturing teams and AI teams need strong foundational skills, there’s no shortcut to building competence.” He adds that, in both arenas, “hands-on practice is irreplaceable,” and context around why work matters is key to excellence.

However, he states, “The pace of change is dramatically different. In manufacturing, core processes might evolve over years or decades. In AI, frameworks and best practices can shift in months.” This divergence requires adaptability and “technological empathy”—understanding both the impact of automation and the value of human capability.

Empirical studies on hybrid deep learning models reinforce the finding that rapid evolution in digital tools necessitates flexible workers who are able to continuously update their skills.

AI-driven learning and virtual reality

AI-powered training tools and immersive simulations are seen by Kikani as catalysts for scalable, personalized workforce development. “Large language models can provide personalized guidance and adapt to individual learning speeds at scale,” he notes, allowing for location-independent learning that would have been impossible just a few years ago.

Virtual reality, meanwhile, “Addresses a different but equally critical challenge: how do you provide hands-on practice for dangerous, expensive, or rare scenarios?” he explains. By enabling repeated safe practice in realistic settings, VR mitigates traditional barriers to high-stakes training, a trend demonstrated by recent expert-reviewed VR platforms in workforce education.

The best implementations maintain strong human involvement, he argues, “As technology scales the mechanics of learning, but humans provide the meaning and motivation.”

Accessibility and benefit for all

Kikani emphasizes the importance of developing AI solutions that are equitable for every employee. “The first principle involves end users from the beginning. What language do they speak? What’s the noise level in their environment? Do they have reliable internet access?” This inclusive approach ensures that tools are designed with lived realities in mind.

He advocates for: “Progressive disclosure—simple, clear interfaces for routine use, with the ability to drill deeper for those who need it.” Explainability is never optional— “When we develop risk assessment frameworks or pattern recognition algorithms, explainability isn’t a nice-to-have feature—it’s a core requirement.”

Incorporating user feedback is ongoing, with continuous measurement and iteration ensuring systems remain relevant and beneficial. These priorities align with current standards for scalable generative AI processes that emphasize governance and accessibility.

Bridging legacy operations and intelligent automation

Looking ahead, Kikani has pragmatic advice for organizations navigating digital transformation. “Start with problems, not technology,” he advises—pinpointing operational bottlenecks before introducing advanced solutions. Cross-functional collaboration is integral, since neither technical nor operational staff alone can deliver effective results.

Kikani stresses the need for “operational discipline,” mirroring his experience with contingency planning in manufacturing settings. “AI systems require the same operational rigor as any critical business system,” he explains.

Metrics for performance must be continually tracked, and interventions readied in case of system failures. These recommendations resonate with best practices for efficient scaling of LLM training, which prioritize accountability in deployment.

“The most important advice I can offer is this: technology is the easy part. The hard part is developing your workforce to work effectively with intelligent automation.” Training and change management are as essential as the underlying technology—only through sustained investment in employees can organizations realize the promise of intelligent automation.

The evolution from traditional operations to intelligent automation is not a simple replacement of roles or a top-down technological mandate. Instead, as Kikani’s trajectory indicates, blending hands-on domain expertise with digital tools creates capabilities that are both resilient and adaptive. The integration of AI into workforce development and operational excellence will draw most powerfully from those leaders and practitioners immersed in both worlds, able to translate lessons from the factory floor into the algorithms that shape the future of work.