agnish
ghosh

Hi, I'm Agnish, a 24-year-old from Kolkata. I currently work as an Ethereum Core Developer on the Nimbus Ethereum Consensus Client, usually called Nimbus-Eth2. It's a lightweight, solo-staker-friendly — or as I'd put it, low-resource staking-friendly — consensus client, written in the Nim programming language. This is the story of how I found my way into Ethereum core development, and into the broader ethos of open-source software engineering for a fair and free internet.

I come from a family of farmers in rural West Bengal, and my parents moved to Kolkata when they were young. By the time I came along, ours was a fairly academic household. My dad taught as a professor at several government medical colleges across West Bengal over the years, while practicing as a physician of Internal Medicine. My mum studied Electronics and Physics, but more than anything she was an avid connoisseur of Indian classical music. Eventually she left her mainstream technical career and took to teaching physics and music at the high school level.

While I was growing up, my dad was mostly away from home, working tirelessly to build his career. So I spent most of my childhood with my mum, wandering the deep rabbitholes of Indian classical vocal music. Before Ethereum, before computer science, before I ever studied math seriously as a high schooler, my curiosity and critical thinking were first ignited by the vast complexities of Indian musical notation.

The twisty, turny notations of music, and the categorization of notes into raagas, taught me how to classify information along many dimensions — emotion, season, time of day, and so much more. The bar and beat combinations, and the discipline of fitting a song within the frame of a tempo, taught me to look at the same thing through different lenses, and to creatively compose a sweet, soothing flow of tone out of thin air. It almost felt like painting a picture with your voice. Only much later did I realize these were my first lessons in Probability, Combinatorics, and Algebraic Group Theory.

Through music and art, I was also introduced to my first open-source communities — long before I knew the term. Across the city and in my own neighbourhood, there were communities where children like me were taught music, painting, dance, and drama for little to no money (perhaps just enough to keep the lights on). Everyone was welcome to learn; if you couldn't pay, that was okay too. That was, honestly, my first exposure to building for the greater good — the holistic nurturing of the young artists of tomorrow.

After school, I began studying Computer Science Engineering at a fairly ordinary college in my city. I did consider pursuing music professionally, but my family didn't have the means to help me break into the music industry — and truthfully, I couldn't stomach the idea of 'selling' an art I had nurtured with so much care and caress. Back then, I had no idea that open-source engineering was even a real career path. For me, like most engineering students around me, landing a job at Google or Amazon seemed like the ultimate endgame of college life.

In college, though, the eerie similarities between Indian music and mathematics resurfaced — this time in my algorithms classes. Solving algorithmic problems felt like a newfound passion, and I found the same joy in designing application backends to be efficient and low-overhead that I had once found in fitting a melody within a tempo. Later, I drifted toward machine learning. This was before LLMs had really entered the picture — in Natural Language Processing, Transformers and BERT were the trending research of the day. I took up research internships at several Indian colleges, but slowly began to notice the gap between academic research and its practical, real-world use. I often found myself demotivated, wondering whether my work would ever see real adoption — whether I could ever make an impact on the world with it.

The lack of funding at several institutes didn't help either. I have vivid memories of holding a table fan in front of my 8 GB college laptop while it strained to train models and churn through data. It was demoralizing — I couldn't see an end to it. On the other side, I had applied to several FAANG companies, only to realize how hard it is to get noticed when you come from an ordinary college. I wanted to pursue the parts of computer science I was passionate about, but I constantly felt like I wasn't doing anything meaningful enough.

Then one day, to my own surprise, a few YouTube videos introduced me to the Rust programming language. As I dug deeper, I stumbled into the world of implementing cryptographic and blockchain protocols with it — Rust being a popular, fast, and efficient choice for that kind of work at the time. A few months later, some internet friends told me about the Ethereum Protocol Fellowship. I applied immediately — and was rejected even faster! But out of sheer excitement, in my third year of college, I decided to participate permissionlessly anyway. After three months of work, the Ethereum Protocol Support team offered to make me a formal, permissioned participant — with compensation. During EPF, I mostly worked on Verkle cryptography in Nim for Constantine, a cryptography library written in Nim largely for the Ethereum core development ecosystem. At the time, Verkle trees were a promising path to shipping Statelessness on the Ethereum roadmap. After six months of EPF, I interviewed with the Nimbus team, and a few months later, I joined Nimbus full-time.

At Nimbus, I initially helped with the stateless effort on Nimbus Eth1, then transitioned to the Nimbus Eth2 consensus client team to work on PeerDAS, which was an upcoming feature at the time. Over the next two years or so, I helped ship Fusaka for Nimbus to mainnet. These days, I mostly work on performance optimizations for the CL client along with general maintenance, and I've recently started working on FOCIL for the Hegota fork — the one that comes after Glamsterdam, the next Ethereum hardfork.

Looking back, none of this would have been possible without the people and institutions that kept the doors open for a kid from Kolkata with an 8 GB laptop and a lot of stubbornness. I owe a deep debt of gratitude to the Ethereum Foundation, for nurturing an ecosystem where anyone, anywhere, can contribute to the core of the protocol; to the Ethereum Protocol Fellowship, for proving that permissionless really means permissionless; and to the entire Nimbus team, for their mentorship, patience, and for trusting me with work that ships to mainnet. The community music schools of my childhood taught me that knowledge shared freely nurtures the artists of tomorrow. Ethereum taught me the same is true for builders — and I hope, through my work, to keep those doors open for whoever comes next.