'Early FSD Energy': Why Top Tesla Talent Is Migrating to Sunday Robotics

The robotics and AI industry is accustomed to talent wars, usually involving massive tech giants trading researchers for exorbitant compensation packages. But a notable transfer of specialized talent is underway that signals a different kind of shift: a cohort of senior engineering leaders leaving Tesla to join the seed-stage startup Sunday Robotics.

Following the startup's emergence from stealth, a report by Electrek highlighted a "brain drain" of engineers moving from the EV giant to the newcomer. In a series of posts on X, those engineers have confirmed their moves, offering a rare glimpse into why top-tier talent is trading the security of a trillion-dollar company for a seed-stage bet.

For these veterans, the draw appears to be a mix of technical philosophy and nostalgia for the high-stakes innovation culture of Tesla’s past.

The "Early Days" Factor

The most telling endorsement came from Nishant Desai, a five-year veteran of Tesla’s machine learning team who worked on Autopilot and Full Self-Driving (FSD). Confirming his move to Sunday Robotics on November 21, Desai drew a direct parallel between the startup and the most productive era of his previous employer.

"This place reminds me so much of the early days of FSD: working with some of the smartest people in the world on some of humanity's hardest problems (and having a blast while doing it)," Desai wrote.

This sentiment suggests that while Tesla scales toward mass production, Sunday Robotics is offering the "zero-to-one" engineering challenges that originally attracted this talent to Elon Musk's orbit.

The Infrastructure Architects

The departures are not entry-level; they represent deep institutional knowledge regarding the AI infrastructure that powers Tesla's current dominance.

Nadeesha Amarasinghe, formerly the Engineering Lead for AI Infrastructure at Tesla, confirmed he has joined Sunday after seven years. His tenure at Tesla covered the critical transition from explicit coding stacks (v9) to the end-to-end neural networks (v14) that now drive millions of vehicles.

"I’ve had the opportunity to build and lead the ML systems that now power FSD and Optimus," Amarasinghe stated, noting that he is leaving to join a "focused, strong and full-stack team" that includes some of his favorite former colleagues.

His reasoning for the move validates Sunday's specific technical bet: "HW/SW/ML co-design." Amarasinghe highlighted that Sunday’s robot, Memo, is "co-optimized" with its data collection and software, rather than being a generic humanoid form factor forced to learn tasks it wasn't designed for.

Breaking the Data Bottleneck

Also confirming his departure was Perry Jia, a six-year Tesla veteran who led Data Engine programs for Optimus and Autopilot. Jia revealed he actually joined the startup in "late June," serving as the Head of Data Operations.

"I’m honored to work with [founders] Tony Zhao, Cheng Chi, and an extraordinary team as we push the boundaries of embodied AI: bringing general-purpose robotics out of the lab and into the home," Jia wrote.

Jia’s move is particularly significant given the strategic divergence in how the two companies approach learning. While Tesla leverages its massive fleet for autonomy and a mix of simulation and teleoperation for humanoid training, Sunday Robotics is attempting to bypass the need for a robot fleet entirely during the training phase.

Instead, the startup utilizes a distributed approach using $200 "Skill Capture Gloves." By distributing these gloves to hundreds of "Memory Developers" in real homes, Sunday aims to crowdsource the "messy" data of domestic life—lighting changes, clutter, and distinct floor plans—without needing to deploy expensive hardware first.

For a data operations specialist like Jia, who spent years refining the pipelines that feed massive neural networks, this method offers a novel solution to the industry's most persistent challenge: acquiring high-quality, real-world manipulation data at scale.

A David vs. Goliath Dynamic

The migration is notable not just for the caliber of talent, but for the scale difference. As Electrek noted, Sunday’s recent $35 million funding round would be a "rounding error" in Tesla’s financials. Yet, the startup has successfully poached a "full stack" of talent, including Jason Peterson, a former recruiter for the Optimus program.

While Tesla promises to produce millions of humanoids to "end poverty," Sunday’s goal is more grounded: a wheeled, efficient home robot that does the dishes. For the engineers who built the backend of Autopilot, that focused mission—and the chance to build it from the ground up—appears to be the new frontier.