He is teaching robots to learn the way a new farmhand does - by watching someone do the job once, then doing it themselves.
Most robots are spoiled. They want a clean warehouse, a flat floor, a barcode taped to every shelf, and a GPS signal that never blinks. Take them outside - into an orchard at dusk, a muddy field after rain, a job site where the map is wrong by Tuesday - and they freeze. Mohan Sivam spends his days in exactly that gap between the lab and the dirt.
He is the founder and CEO of Neuralzome Cybernetics, a Bengaluru deeptech company building autonomous robots for off-road work. The premise is almost suspiciously simple. Today, getting a robot to do a real job means hiring people who can write the code for every turn, every obstacle, every edge case. Sivam's answer is to delete that requirement. His robots are meant to be teachable - you show one how to do a task, the way you would show a new hire, and it picks it up. No specialist required.
His flagship is NeuralPilot, an agentic, multimodal autonomy stack designed for the places navigation usually dies: GPS-denied terrain, changing light, ground that shifts under the wheels. It fuses camera, LiDAR and GPS into a single off-road brain, learns from human demonstration, and adapts to terrain in real time. Underneath sits a three-tier safety system - Vision AI, GPS and SLAM - so that when one sense fails, two others keep watch.
In August 2025, the bet got real money behind it. Neuralzome closed a $2.4 million pre-seed round led by 8X Ventures, with Turbostart, Avinya Ventures, Saka Ventures, Appreciate Capital, Astir Ventures, IIM-Ahmedabad's CIIE incubator, SIDBI and angel Heston Castelino joining in. Total raised to date sits near $2.65 million. The plan: push a Robot-as-a-Service model into the high-value, labor-short corners of the economy - orchard management, commercial landscaping, precision agriculture, off-road industrial work.
Ask him to summarize, and he keeps it short: “We are developing technologies that break the critical barriers preventing the adoption of robots in real-world applications.” His LinkedIn tagline is shorter still, and a touch cheeky for a hardware founder - making robots sentient.
We didn't set out to disrupt anything. We just couldn't stop asking why the hard things stayed hard.— The Neuralzome founding philosophy
Neuralzome doesn't ship model numbers. It ships characters. The products carry names that sound like they belong in a story - and one is a wink at Tamil cinema's most famous robot.
An agentic, multimodal autonomy AI for off-road and GPS-denied terrain. Learns from human demonstration; handles perception, navigation, mission execution and live monitoring.
A simulation engine that spins up photorealistic digital twins, so a robot can rehearse a difficult field hundreds of times before it ever touches real soil. Less on-field training, lower cost.
An open-source autonomous outdoor platform and R&D testbed for NeuralPilot. Built to haul a 500-pound payload across rough ground for eight hours at a time.
An autonomous off-road ATV that bolts onto standard vehicles for soil sensing, material transport and precision agriculture. The name nods to the android from the film Enthiran.
Sivam did not arrive at autonomous farm robots in a straight line. He trained as an engineer at Anna University in electrical, electronics and communications, then spent more than a decade learning the unglamorous middle of hard tech - computer vision, graphics, robotics, the connective tissue that makes machines understand the world.
In 2013 he started GAMASOME and ran it as founder and CEO for the better part of eight years. He took a consulting turn at Hero MotoCorp, India's two-wheeler giant, where he led connected-vehicle and charging-infrastructure programs on a global scale - learning, perhaps, how big industry actually adopts new technology, and how slowly. He co-founded Flo Mobility as CTO, working the autonomy problem from another angle. In 2018 he detoured entirely, taking a crash course at MIT on sequential genomics using machine-learning models - robotics' favorite tools pointed at biology.
Then, in 2023, came Neuralzome. When he told his story on the SRX robotics podcast, he framed the through-line plainly: a journey that begins with networking and persistence. Not genius. Not luck. Showing up, and refusing to let go of a problem.
He runs the company alongside co-founders Aditya Shriwastava (CTO, an AI and industrial-robotics specialist) and Prasanna Venkatesan (COO). The split lets Sivam keep his hands on the part he clearly loves most - the question of how a machine can be taught.
We are developing technologies that break the critical barriers preventing the adoption of robots in real-world applications.— Mohan Sivam, on the Neuralzome pre-seed raise
His LinkedIn headline doesn't say “CEO.” It says “Making robots sentient.”
One Neuralzome robot is named Chitti - the same name as the android hero of the Tamil sci-fi blockbuster Enthiran.
His full given name is Paramasivam Mohankumar. The world just calls him Mohan Sivam.
He once studied genomics at MIT - the machine-learning toolkit, aimed at DNA instead of dirt.
Before farm robots, he helped figure out how to charge and connect two-wheelers at Hero MotoCorp.
The training engine is called RedPill - the robots literally wake up inside a simulated reality first.
Sivam's ambition isn't a flashier robot. It's a robot that ordinary people can put to work without a manual. If a grower can teach a machine to mow an orchard the way they'd train a seasonal worker, the economics of automation stop belonging to the few companies that can afford robotics teams.
That is the wager underneath NeuralPilot, RedPill, Jeeno and Chitti: that the real bottleneck in robotics was never the wheels or the cameras. It was the teaching. Solve the teaching, and the hard things finally stop being hard - which, by his own account, is the only reason he started.
If “making robots sentient” made you look twice, send it to someone who still thinks robots only work indoors.