Prasad's Boston Rolodex: 3 Pools Boston Dynamics Will Mine First

On October 7, 2026, Boston Dynamics named Rohit Prasad as CEO and did something recruiting teams almost never do in a press release: it named the talent pools he would raid. The company said his 25-year Boston network "strengthens Boston Dynamics' ability to attract top talent." Translation for every competing sourcer: you have maybe 90 days before his internal recruiters work the same list.

This is a map of that list. Who he knows, where they sit, and the order a sane competitor should hit them in.

Why the Oct 7 announcement is a dated starting gun

Boston Dynamics explicitly told the market it will use Prasad's personal network to recruit, which means the pools are already identified and the race is on. Three of them are named or trivially inferable from his resume: Amazon Alexa and Nova, Raytheon BBN Technologies, and the Boston academic AI and robotics labs he has worked with for a quarter century.

The urgency behind the hiring push is Hyundai, which took majority ownership in 2021 and has publicly committed to deploying Atlas at its Georgia plant from 2028 with capacity for up to 30,000 humanoids a year. That production target implies a multi-hundred-hire physical-AI ramp over the next 24 months. Prasad does not have the luxury of a slow build.

A few things sourcers should internalize before touching a sequence:

• Boston Dynamics HQ is in Waltham, MA. Prasad will not want to lose candidates to relocation, so he is effectively fishing in a 20-mile radius between Waltham, Cambridge, and Route 128.
• He has a documented pattern of re-staffing new orgs from Alexa alumni. Fortune reported his post-ChatGPT AGI mandate at Amazon was "led almost entirely by ex-Alexa executives." He will do it again.
• Outgoing CEO Robert Playter left in February, which means his direct reports have spent eight months in limbo. They are flight risks today, not next quarter.

Pool 1: Amazon Alexa and Nova, Cambridge cohort

The highest-conviction pool is the Amazon Cambridge, MA office in Kendall Square, which housed the original Alexa speech science team and sits two subway stops from Boston Dynamics' recruiting coffee shops. This is not the Seattle Alexa org. The sourcing target is "Amazon + Cambridge MA + speech or NLP or ML," and it is a materially smaller, more poachable set than a national Alexa search returns.

Prasad spent 12 years at Amazon and played a significant role in creating the Amazon Nova foundation model family. Nova is the specific sub-team to mine, because it is where he spent his final mandate and where his trust network is freshest.

The macro condition helps. Over the past two years, Alexa AI and the AGI reboot have been publicly flailing; leaked internal documents cited by Fortune last year identified critical flaws in the delayed Alexa rebuild. Cold outbound into a demoralized org converts at roughly 3-5x baseline. That multiplier evaporates the moment Prasad's retention offers land.

8,170

The practical sourcing move is to narrow by geography and by team keyword (Nova, Lex, Polly, Rekognition, Titan, speech science) before you touch a sequence. In Refolk, the query that works is literally "ex-Amazon Alexa or Nova engineers within 30 miles of Waltham, MA, with speech or LLM experience," written in plain English. You get back a ranked shortlist instead of an 8,000-row CSV to dedupe.

Pool 2: Raytheon BBN alumni, the sleeper pool

BBN alumni are the quietly correct first move, because Prasad spent nearly 14 years there (longer than his Amazon stint) and nobody else is sourcing them. Raytheon BBN Technologies, headquartered in Cambridge two blocks from MIT, is where Prasad led machine learning research for U.S. government and commercial customers before Amazon.

The BBN diaspora has four properties that make it unusually easy to close:

1. Older and cheaper than FAANG ML talent, with equity expectations calibrated to defense contracting.
2. Security-cleared, which matters for the DoD-adjacent work Boston Dynamics does with Spot and future Atlas variants.
3. Loyal to Prasad personally from the BBN years. The intro email writes itself.
4. Sitting in low-visibility employers: Raytheon, MIT Lincoln Laboratory, Nuance (now Microsoft), SRI, and a long tail of DARPA-funded startups most LinkedIn Recruiter seats never surface.

The mechanism is simple: Raytheon BBN alumni sourcing is unfashionable, so the brand reads as "defense contractor" to recruiters chasing GenAI, and the profiles get skipped. Prasad knows these people by first name. A headhunter who gets to them in October has a two-week window before he does.

Pool 3: Boston academic AI and robotics labs

The academic feeder channel is where Prasad will fill entry-level and new-grad roles through faculty relationships, not through career fairs. The press release's phrase "academic affiliations" is doing specific work: it points at MIT CSAIL, Harvard SEAS, Northeastern Khoury, and Brown.

In Refolk's index of professional profiles, Greater Boston and Cambridge both appear in the top ten US regions for robotics-skilled talent, which is unusual density. The practical read is that Prasad can staff without relocation packages, and competitors need to either match the geography or offer remote-first roles as a structural counter.

The named reverse-poach lane is the RAI Institute, Marc Raibert's post-Boston-Dynamics lab. Raibert founded Boston Dynamics, left, and started RAI in the same city with many of the same people. RAI shows up in Refolk's top ten employers for robotics-skilled profiles, alongside Ambi Robotics, GrayMatter Robotics, and Slip Robotics. Every single RAI engineer is a boomerang candidate for either side of this fight.

The numbers side by side

The spine of the pipeline decision is how these pools compare in size and reachability. Refolk's index gives us order-of-magnitude counts to sequence outreach against.

The ratio row is the one worth staring at. Nationally, the robotics-skilled bench is only about 1.3 times the Alexa-keyword pool, which means Prasad's Amazon network is a surprisingly large fraction of all the talent anyone in physical AI is chasing.

The non-obvious insight: Prasad will under-index on manipulation

The gap competitors can exploit is that Alexa and Nova alumni are overwhelmingly NLP, dialog, and foundation-model people, not sim-to-real or robotic manipulation specialists, so Prasad's natural rolodex solves the wrong half of Hyundai's problem.

Humanoid production at the 30,000-unit scale needs manipulation, contact-rich control, teleoperation data pipelines, and sim-to-real transfer. That specialty lives in a different diaspora: ex-Google Robotics, ex-Everyday Robots, ex-Covariant, ex-Dexterity, Berkeley BAIR, CMU RI, Stanford IRIS, and the RAI Institute. Prasad's network does not deeply reach into those labs.

The arbitrage for a competing recruiter or a competing lab is to lock up the manipulation specialists before Boston Dynamics figures out it needs them. The Alexa poach gets the headlines; the manipulation poach wins the actual production ramp.

Prasad's network solves the wrong half of Hyundai's problem, and the manipulation specialists are the arbitrage.

A 90-day sourcing sequence

The ordering that respects Prasad's own ordering is: hit his most-personal relationships first, because those are the ones he will close fastest once his desk is set up. A workable 90-day plan:

1. Days 0 to 14: Playter's former direct reports and the Boston Dynamics VP bench. Eight months of CEO limbo makes them answer the phone.
2. Days 7 to 30: RAI Institute engineers and research scientists. Boomerang risk is the lever; frame around either returning to Boston Dynamics or defending against being pulled back.
3. Days 14 to 45: Raytheon BBN alumni now at Raytheon, MIT Lincoln Lab, Nuance/Microsoft, and SRI. Low-competition, high personal-loyalty to Prasad.
4. Days 21 to 60: Amazon Cambridge Alexa and Nova engineers. The Fortune narrative about Alexa flailing is your opener.
5. Days 30 to 75: Senior PhDs and postdocs at MIT CSAIL, Harvard SEAS, Northeastern Khoury. Faculty intros matter more than InMails.
6. Days 45 to 90: Manipulation and sim-to-real specialists outside Boston. This is the pool Prasad will reach for late, which is your window.

The reason to run these in parallel instead of serial is that steps 1 through 3 are shared with Boston Dynamics itself, so you are racing. Steps 5 and 6 are shared with Figure, 1X, and Physical Intelligence, so you are racing a different set. Different sequences, different pitches, different week.

For teams running this across 500 to 2,000 target profiles, the bottleneck is almost never the search; it is dedup, enrichment, and keeping the Cambridge cohort separate from the Seattle one.

What a competing lab's pitch should sound like

The counter-offer that beats Boston Dynamics is not more cash; it is either geography (remote, SF, or Bay Area in person) or scope (manipulation and embodied AI, not voice interfaces again). Prasad's structural weakness is that his comfort zone is Waltham-Cambridge and his comfort stack is NLP and foundation models. Anything outside that is a fair pitch.

Three framings that work on his likely targets:

• To Alexa alumni: "You already shipped the voice assistant. Do the physical one somewhere that is not a reboot of a reboot."
• To BBN alumni: "Same security-cleared mission, same Cambridge commute, equity that vests before the IPO."
• To academic hires: "A research charter, not a product deadline tied to a Hyundai plant opening in 2028."

Boston Dynamics hiring at this scale is a market event, not just a company event. Treat it that way.

FAQ

How urgent is the window on ex-Amazon Alexa engineers?

Roughly 90 days from the Oct 7 announcement before Prasad's internal recruiters have their outreach calibrated and retention conversations have happened on the Amazon side. The leaked documents around the Alexa reboot and the broader AGI struggles mean cold outbound converts at an unusual multiple right now, but that softness closes the moment Amazon counters with equity refreshes or Prasad personally calls his old reports.

Why focus on Raytheon BBN alumni when the brand is so dim?

Because dim brands are where competition is lowest and personal loyalty is highest. Prasad spent nearly 14 years at BBN, longer than at Amazon, and the alumni network is largely unlisted on sourcing tools because recruiters chase GenAI brands instead. BBN alumni are typically security-cleared, Cambridge-local, and reachable through warm intros, which makes them the single highest-yield sleeper pool for physical AI recruiting in Boston right now.

Does Boston Dynamics really need hundreds of new hires?

Hyundai's publicly stated capacity target is up to 30,000 humanoids a year starting from the Georgia plant in 2028, which is a capacity figure rather than a shipment number but implies a multi-hundred-person hiring curve over 24 months to design, train, and support the fleet. Not all of that lands in Waltham. Expect a split between Boston (research, software, controls) and Georgia or Korea (manufacturing, deployment, field engineering).

What is the single highest-conviction name to source first?

A current RAI Institute engineer who was at Boston Dynamics pre-2022. They have the exact domain fit, a personal relationship with Marc Raibert, and now a second Boston robotics mandate under a well-known CEO one commute away. Whichever competing lab reaches them first in October will set the tone for the rest of the quarter.