Prathamesh Joshi

I’m a Co-founder & Principal Engineer at Carnot Technologies, now part of the Mahindra Group. For most of the last decade I’ve worked with our teams on IoT and telematics — the devices, the platform behind them, and most of what sits in between.

I prefer work with visible physical impact: plug-and-play IoT for farmers, where we found genuine product-market fit, and connected systems for urban mobility. When the output is a tractor that runs better or a fleet that stops quietly losing fuel, you find out fairly quickly whether the engineering was any good.

Most of it happens where networks are unreliable and edge behaviour is imperfect, so over time I’ve come to care more about reliability, security and operational simplicity than about anything clever.

• Build full-stack systems: device and firmware, gateway, backend, and the data and analytics on top.
• Get hardware into the real world: aftermarket retrofit kits and factory-fit OEM integrations — CAN bus, immobilisers, BLE sensors — plus the manufacturing, SIM and installation tooling a hardware fleet quietly runs on.
• Design for failure: retries, partial failures, bursty traffic, and telemetry that lies.
• Operate cloud platforms: Kubernetes and infrastructure as code across AWS, GCP and Azure, plus the security and cost work that keeps a large fleet affordable.
• Build for the people actually using it: apps for farmers, operators and dealers — often first-smartphone users, in several languages, on networks that come and go.

Much of my waking life over the last decade has revolved around Carnot. Nothing I can claim to have built there was built alone — it was co-created with my co-founder Pushkar and a team at Carnot who were equal partners in all of it. Our philosophy as a compact, agile team of thinkers and doers is well captured in this post by Pushkar, and a quick tour of what we’ve built lives in this flipbook brochure.

Carnot itself has been a rollercoaster. We started in 2015 on car telematics — a plug-and-play dongle that kept an eye on your car when you were nowhere near it — theft and tow alerts first, then diagnostics, and a score for how you had been driving. Winning Qualcomm’s Design in India Challenge in 2016 — with a platform that made any ordinary two-wheeler smart — moved us into urban mobility and securing bikes for shared-mobility fleets. Then Mahindra & Mahindra came in as a strategic investor in 2018, and agriculture turned out to be where the idea actually belonged. M&M holds a majority stake today and Carnot is a Mahindra Group company.

Agritech is where the company found what it was for. Carnot now runs one of the largest agri-fleet telematics deployments in India, and we are still connecting farm machinery to the platform every day. Getting there took rather more wrong turns than that summary suggests.

In the early days in Carnot, when we custom-built our own hardware, I was deep in firmware and embedded systems. As the needs changed, I moved to whatever helped most: native mobile apps (Android and iOS), backend services, and running the hosted infrastructure behind it all — a decade of commits traces how that shifted year to year. A lot of this happened in the pre-GPT era, which meant burnt hands. Also a few too many late nights, and the occasional detour into website and Unity work to put out fires (and try not to start new ones). Six granted patents came out of that stretch, on things like fuel consumption monitoring, GPS trajectory correction and area estimation from geospatial data — all of them shared with the people I built them with.

Before Carnot, my master’s thesis focused on metal Additive Manufacturing, done in collaboration with Sajan Kapil, and the work still excites me to this day. It produced a dozen or so papers on hybrid-layered manufacturing, slicing strategies and support-free printing — which remain, slightly to my amusement, the most cited things I have ever written — along with two more granted patents.

Even earlier, during my undergraduate years, I was part of IIT Bombay Racing in its formative phase. I was one of the firmware authors on the first two cars we raced at Silverstone, writing the electronic differential and the on-board data acquisition system for Evo 2.0 — the car that made us the first Indian team to qualify for the endurance race at Formula Student Silverstone in 2013. Once again, alongside Pushkar. The team has since gone on to achieve remarkable success, a testament to the generations of engineers that followed.

• LinkedIn
• GitHub
• Carnot
• A decade of commits
• Google Scholar
• Email