About
Product and GTM leader. Most recently CEO of an AI startup Yuuki, previously Group CPO at…
Articles by Nihit
-
My reflections and update on my most potent life journey yet
My reflections and update on my most potent life journey yet
Who we are, if not a collage of our most eccentric and strongly held opinions. The other day I got into a rather…
15 Comments -
Ratings need a makeover, why and how of it!
Ratings need a makeover, why and how of it!
Godfather is rated 9.2/10 on IMDB.
4 Comments
Activity
6K followers
-
Nihit Nirmal shared thisThe new BS in town is: “taste is the new moat.” Then people who say that go on to build a podcast summarizer. The point of an information stream isn’t to get a cleaner summary. It’s to sharpen your point of view. And that point of view should feed a real decision. But when you ask an LLM for help, you often get the opposite: more surface area, less conviction. Especially when the task is “summarize this article” or “give me the takeaways.” I’ve been thinking about how to turn what I’m reading and writing into better decisions, so I built a small tool for myself. At any point, I have 3-4 active decisions. For each one, I keep a working folder with my current stance and key questions I am grappling with. The pipeline is, ingest what I’m reading, map each chunk against an active stance, retrieve related notes from my own writing, and synthesize where the real leverage point is. So when a new article comes in, rather than asking "what is this about?” It asks: Does this strengthen my view, challenge it, or expose a gap in it? Then it checks my own notes for where I’ve already been circling the same idea, but hadn’t connected it clearly. The job isn't summarizing the content. It's testing my thinking against it. I’ll share a walkthrough next week: how the mapping works, how it connects to prior notes, and what happens when the model gets it wrong. The fundamentals of learning haven’t changed. We still need to read, process, and develop our own thinking. What has changed is the amount of noise.
-
Nihit Nirmal shared thisAs we started building with LLMs at Yuuki, like any other product person, I brought in the workflow that worked for the last 15 years. Spec it, build it, test it, ship it. The assumption was the same: you know what good looks like before you start testing. So we built evals to do exactly that. Score the output, check the quality, move on. However as LLM outputs are non-deterministic, 10 point evals created problems. My nine turned out to be my co-founder's six. 😆 The LLM judge scored a seven that matched neither of us. Two months of scoring and we couldn't tell if the product was improving or our standards were drifting. You can't test for correctness when correctness isn't fixed. A coaching response that's empathetic but doesn't push the user to act, is that good? Depends on who you ask, depends on when you ask them. Hamel Husain's work on LLM evaluation pointed us to the fix. Binary. Yes or no. Is this response empathetic? Yes or no. No more debates about sixes and sevens. The feedback loop when building with LLMs is extremely tight. Every binary judgment came with a line of commentary. "Gave advice before asking a single question", that critique becomes a constraint. The next version follows it. Gets reviewed again, refined further. The evals aren't just a testing phase. They're the loop through which the product gets defined. And now because everyone, product, engineering, domain experts, is judging the same outputs, the evals force alignment on what good actually means. That alignment didn't exist before. The process created it. More on this in the next post. For now, wrote a detailed post on how we designed our evals, link in comments.
-
Nihit Nirmal shared thisFunny episode of the day: I was chatting with Claude, which has a nice sense of humor. I was supposed to say “super helpful", and Claude's response was so on point, that I ended up replying to a non-required response.
-
Nihit Nirmal posted thisEverywhere I look, teams are celebrating AI coding agents. Million lines generated. PRs flying. Shipping velocity through the roof. Having lived through this, let me share what it actually looks like on the other side. After 24 months of building Yuuki and now contributing to Rapida, there is a teeny tiny reality that changes how agentic engineering works. Agents are like engineers: they write a lot of code. But they are unlike engineers because they don't carry context. They start fresh every time. The code is technically correct but contextually naive. For the first feature, that's manageable. By the tenth deploy, you're shipping code that ignores everything production taught you from the first nine. That's not speed. That's compounding tech debt at machine pace. One of our agents at Rapida generated a webhook handler that passed every test. In production, the telephony provider retried on slow responses and the handler wasn't idempotent, which led to duplicate calls. An engineer who'd been bitten by that provider before would have built idempotency in from the start. The agent didn't have that scar tissue. Because of our traces, we caught it within minutes and fixed it the same day. Then we encoded the lesson, idempotency constraints baked into the webhook contract. Now agents generating webhook handlers must follow that contract. The system carries the context so the agent doesn't have to. That's the pattern: not more reviews before shipping, but a feedback loop where production teaches the system, and the system constrains the next generation. The teams shipping reliably at AI speed aren't generating the most code. They're the ones whose system learns from every deploy. Link to the blog post in the first comment.
-
Nihit Nirmal shared thisPrashant Srivastav and I had a fight during our early days of building Yuuki. I'd overcommitted — three enterprise clients, four conflicting requirement sets, ten days to ship. Fujifilm couldn't store any data. TPC Group wanted anonymized conversations but no recordings. SMU needed anonymous access for student test groups. I told him we had to do it. He told me I was over-promising. I knew he was right. He knew we'd have to do it anyway. That fight wasn't really about those three clients. It was about what we were signing up for every time we said yes, absorbing custom enterprise logic into a single codebase with no QA team, no SRE, no one to catch what we missed. In the "10x engineer" era, we're told small teams can do more than ever. Nobody talks about the cost of "more." GitClear's 2024 study found that while AI adoption is skyrocketing, code churn — code pushed and immediately reverted or fixed — is expected to double. We're generating AI code faster than we can verify its quality. We started calling this the Quality Gap. The delta between how fast you can ship a feature and how long it takes a client to actually trust your code in production. Most teams try to bridge it with feature flags. But we weren't toggling UI buttons — we were toggling core infrastructure. Custom ERP connectors. PII masking logic. With 10 clients on 10 custom branches, you're looking at 1,024 possible system configurations. That's not a testing challenge. That's a testing explosion. So we built something different: Virtual Isolation. Three separate client repositories, each with its own identity. Before code hits a client repo, it runs isolation checks — if Fujifilm's logic touches a global variable it shouldn't, the build fails. Before client repos merge into the core, they run contract gates to verify the API schema is untouched. The whole thing syncs automatically: core pushes plumbing out, spokes push approved logic in. We used to spend 14 days in email hell for every UAT cycle. Now we're under 48 hours. In Over 20 months that's the thing that actually scaled. We went from shipping fast to building a discipline layer that makes speed safe. You don't need a bigger team. You need a tighter gate. If you're scaling enterprise SaaS with a tiny team, we wrote up the full hub-and-spoke architecture. Link in comments.
-
Nihit Nirmal posted thisThere's simplicity in using one AI model for everything. Having said that, I've found there's a lot more value in building an ensemble. Last night I was putting together an evaluation framework for a Return and Refund agents for an e-commerce company, basically trying to define what "good" looks like when a system makes autonomous decisions on behalf of your customers. Accuracy, autonomy, escalation, drift. The usual suspects. I started with Gemini to do the research. It's weirdly good at finding stuff that doesn't surface in normal search — obscure papers, niche technical podcasts, the kind of references you'd only get from someone deep in the field. Then I brought all of that into Claude and started poking at it. "Is false positive rate actually the right way to measure agent accuracy?" And that's where things got interesting. The dangerous metric isn't wrong answers — it's wrong answers the agent was confident about. A low-confidence mistake gets flagged and routed to a human. Fine. A high-confidence mistake? That one executes on its own. - That's the one that costs you money. Gemini found the raw material. Claude was the sparring partner that pressure-tested it, and helped me get there in an hour vs a couple of days. But but but, connecting how metrics interact in production, finding what's missing between them — that's a different kind of thinking. So I used Manus to stress-test the full picture. That's when the few more gaps showed up — things like whether the agent's mistakes are getting worse over time, or whether it handles irreversible actions (processing a refund) differently from reversible ones (looking up an order). I think what's actually going on, and my view on this changes every hour, lol — is that every model has a fixed thinking budget. Some spend it sorting through messy inputs. Others spend it on complex reasoning. When you ask one model to do both, quality falls apart. The 10X isn't a smarter model. It's a clearer question: what kind of uncertainty am I actually reducing right now?
-
Nihit Nirmal shared thisBuilding an AI tool takes a weekend. Making it survive Monday morning takes months. I've been building something I use every day — AI agent workflows that sit on top of everything I read and write. It coaches me through daily execution and weekly reflections. It knows my quarterly goals, what stresses me out, and how I handle it. My stack: Obsidian, Claude Code, and qmd for vector search across 10,000+ articles /books I have read over past 7 years pushed via Readwise. Building this was easy. Keeping it alive is where it gets interesting. Just the other day, I was building a presentation deck with a critic agent evaluating each slide and the whole deck. Everything was passing. Then vector search surfaced Nancy Duarte's work from my readwise library — titles should be assertions, not labels. One idea per slide. I rebuilt the eval around her framework. Suddenly every slide failed. Same slides, better lens. That's the feedback loop — build, evaluate, have your own knowledge base sharpen the evaluation, rebuild. It only exists when you're using the system for real work. Demos don't give you this. Three months in, the coaching agent holds me accountable across weeks helped me log 50 hours of 'deep & consequential work' every week It doesn't forget what I committed to on Monday when I try to skip it on Thursday. Skip the demo. Build something that has to work tomorrow morning. The feedback loop is the product.
-
Nihit Nirmal shared thisWe spent two years building voice AI infrastructure that worked in production. Sub-1.2s latency, 1000+ concurrent calls, ran systems for Fujifilm, TPC Group, Meesho etc. Rather than let that work disappear while other teams rebuild the same stack from scratch, we open-sourced it as Rapida. Rapida has picked up 500 stars since launch last 2 months. We released Remote Agent Executor, MCP integration just last month. Prashant Srivastav has been contributing heavily to the effort. Too many voice AI platforms lock you in. You share your data with them to solve orchestration, get locked into their roadmap, their pricing, their stack. Rapida gives teams control. Your infrastructure, your data production-ready. We're also planning to open-source more of the underlying systems, the evals, the synthetic data creation, the quality tooling that made it work at scale. If you're building with voice AI, worth a look, https://lnkd.in/giwEMFG7 DMs open if you want to talk through your stack.Nihit Nirmal shared thisThree releases in one month. We open sourced Rapida last month with a clear commitment to move fast and ship real features. What we shipped: Remote Agent Executor Run your voice agents anywhere using any LLM, framework, or language. Your cloud or servers. Agent builders focus on logic while infra teams control runtime and deployment. Model Context Protocol (MCP) Your voice agents can now talk directly to business systems like Salesforce, HubSpot, ERPs, and your own APIs. No middleware complexity. Just integrate and deploy. Most voice AI platforms today are closed, rigid, and opinionated. They move slowly, hide complexity, and lock teams into their stack. In just one month of open development, Rapida has become richer and more flexible than many closed source platforms. Transparency and real user feedback beat black boxes every time. That’s what we’re building. Every week. Every commit. If you’re evaluating voice AI platforms or thinking about building something custom, give Rapida a look. It’s open source, production ready, and built for teams running real systems. Repo: https://lnkd.in/gqhX6RHN Release notes: https://lnkd.in/gJjUxzMA DM me if you’re looking to host Rapida on your private cloud. #VoiceAI #OpenSource #AI #AgentFramework #DevTools
-
Nihit Nirmal shared thisI shut down Yuuki last month. Returned capital rather than drag it out. I'm supposed to write the gratitude post now. The journey was the destination, I'd do it all again. I'm not going to write that post. When I started Yuuki, I believed AI could make coaching accessible at scale. The conventional wisdom in late 2023, was that it was too early, too hard, too niche. So my job became simple: run the experiments, get signal, decide. We ran three experiment categories and their outcomes 1. 𝗖𝗮𝗻 𝘄𝗲 𝗯𝘂𝗶𝗹𝗱 𝗶𝘁 𝘄𝗲𝗹𝗹? Yes, blind evals for daily work situations showed parity with human coaches. 2. 𝗪𝗼𝘂𝗹𝗱 𝗲𝗻𝘁𝗲𝗿𝗽𝗿𝗶𝘀𝗲𝘀 𝗽𝗮𝘆? - Yes, we got Fujifilm, TPC Group, Meesho as paying customers. 3. 𝗪𝗼𝘂𝗹𝗱 𝘂𝘀𝗲𝗿𝘀 𝗮𝗰𝘁𝘂𝗮𝗹𝗹𝘆 𝘂𝘀𝗲 𝗶𝘁? - No. Activation was fine, but engagement dropped after two weeks. Users wanted resolution, not more reflection. Two experiments cleared. The third didn't. We iterated hard - fixed our evals to fine tune flywheel, reduced number of turns to give resolution, added the voice mode, brought in privacy controls. Each change made the experience better, but did not bend the engagement curve. 𝗜𝗳 𝘆𝗼𝘂𝗿 𝗽𝗿𝗼𝗱𝘂𝗰𝘁'𝘀 𝘃𝗮𝗹𝘂𝗲 𝗼𝗻𝗹𝘆 𝘀𝗵𝗼𝘄𝘀 𝘂𝗽 𝗮𝗳𝘁𝗲𝗿 𝗮 𝗻𝗲𝘄 𝗵𝗮𝗯𝗶𝘁 𝗳𝗼𝗿𝗺𝘀, 𝘆𝗼𝘂𝗿 𝗿𝗲𝗮𝗹 𝗰𝗼𝗺𝗽𝗲𝘁𝗶𝘁𝗼𝗿 𝗶𝘀𝗻'𝘁 𝗮𝗻𝗼𝘁𝗵𝗲𝗿 𝘀𝘁𝗮𝗿𝘁𝘂𝗽. 𝗜𝘁'𝘀 𝘁𝗵𝗲 𝗱𝗲𝗳𝗮𝘂𝗹𝘁 𝗱𝗮𝘆. 𝗧𝗵𝗲 𝗺𝗲𝗲𝘁𝗶𝗻𝗴𝘀, 𝘁𝗵𝗲 𝗦𝗹𝗮𝗰𝗸, 𝘁𝗵𝗲 '𝗜'𝗹𝗹 𝗱𝗼 𝗶𝘁 𝗹𝗮𝘁𝗲𝗿' 𝗽𝗶𝗹𝗲. 𝗔𝗻𝗱 𝗳𝗼𝗿 𝘂𝘀, 𝘁𝗵𝗲 𝗱𝗲𝗳𝗮𝘂𝗹𝘁 𝗱𝗮𝘆 𝗿𝗲𝗺𝗮𝗶𝗻𝗲𝗱 𝘂𝗻𝗱𝗲𝗳𝗲𝗮𝘁𝗲𝗱. 😒 Some problems don't compress on your timelines. Not because you didn't work hard enough, but because the problem is shaped that way. Behavior change at scale is one of them. Once we had that answer, continuing would have been protecting my ego, not the company. When I started Yuuki, I was a product person who needed someone else to code and someone else to sell. Two years later, I do both, and the version of me that exists now wouldn't have existed without these two years. I used to think sales was performative. Turns out the good version is the opposite, I learned to sell by sitting with buyers long enough to hear what they weren't saying. I learned to build by shipping things that mattered to me when the company was falling apart. An AI chief of staff I still use every day. 𝗦𝗼𝗺𝗲𝘄𝗵𝗲𝗿𝗲 𝗮𝗹𝗼𝗻𝗴 𝘁𝗵𝗲 𝘄𝗮𝘆 𝗜 𝗿𝗲𝗮𝗹𝗶𝘇𝗲𝗱: 𝘁𝗵𝗲 𝗳𝗼𝘂𝗻𝗱𝗲𝗿 𝗷𝗼𝗯 𝗶𝘀 𝗹𝗲𝘀𝘀 "𝘃𝗶𝘀𝗶𝗼𝗻𝗮𝗿𝘆" 𝗮𝗻𝗱 𝗺𝗼𝗿𝗲 "𝗰𝗼𝗺𝗽𝘂𝗹𝘀𝗶𝘃𝗲 𝗹𝗼𝗼𝗽-𝗰𝗹𝗼𝘀𝗲𝗿." 𝗛𝗲𝗮𝗿 𝗽𝗮𝗶𝗻. 𝗕𝘂𝗶𝗹𝗱. 𝗦𝗲𝗹𝗹. 𝗟𝗲𝗮𝗿𝗻. 𝗥𝗲𝗽𝗲𝗮𝘁. When shutting down felt like failure, building became the thing that kept me moving. I'm deep in the AI + enterprise world now, building sales automation tools. If that's your world, would love to talk.
Nihit Nirmal commented on a post
1w
This is so good.. Now I can see why you were so excited about this trip :-)
-
Nihit Nirmal liked thisNihit Nirmal liked thisA key benefit of recent technological advancement has been the ability to scale and democratise access to opportunities that were once limited to the privileged few. We have partnered with EZRA over the past few years to bring top quality professional coaching to our talented colleagues at Standard Chartered and it's been great to see the collective progress. I was delighted to join Nick Goldberg, his fantastic Ezra team, and a brilliant group of global talent executives in Rome this week to discuss the implications of AI, and other trends, for how we continue to develop the next generation of leaders, and the continued access to coaching (both Human and AI) that is key to accelerating their progress. Thanks to Ezra for the opportunity to shape the conversation, and to my fellow participants for an enriching dialogue. Sinead Keenan Susanna Kable Tim MacCartney Hillary Champion Haesun Moon Brian Murphy Svenja Carolin Dietrich Vidya Krishnan John J. Kelly Jr. Helen Basford PCC
-
Nihit Nirmal liked thisNihit Nirmal liked thisWe’re building at an exceptional pace at Emergent, one of the fastest-growing AI startups today. And at this stage, every decision we make about people - who we hire, how we enable them, and how we reward them - directly shapes the company we become. I’m looking to hire an HR leader who will work closely with me to build this, intentionally. This role sits at the intersection of people, performance, and business outcomes. It’s about deeply understanding people: how they think, what drives them, and what unlocks their best work. And equally, it’s about first-principles thinking and a business-first mindset. Because HR here is not a support function. It is how we scale excellence. You will partner with founders and leaders to: – build an environment where ambitious, high-agency people thrive – ensure individuals are focused on what truly moves the needle – design systems where performance and rewards are meaningfully aligned – make clear, high-quality calls on people and team decisions A few things this role is not: – it’s not about maintaining processes – it’s not about optimizing for comfort – it’s not about playing it safe This is a role for someone who wants to shape the company, not just support it. If you care deeply about people, think independently, and want to build an organisation where great individuals do the best work of their lives, we should talk. If you or someone you know is the right fit, write to me at talent@emergent.sh with a note on why you are the best person for this role. I look forward to reading some great responses.
-
Nihit Nirmal liked thisNihit Nirmal liked thisGreat discussion at Money20/20 on the future of cross-border payments. As consumers increasingly live, travel, and transact globally, their expectations have changed. I enjoyed discussing how the industry is coming together to solve these challenges through stronger local payment rails, embedded wallet experience, and use of AI. Thank you for having me on the Radiant Stage.
-
Nihit Nirmal reacted on thisNihit Nirmal reacted on thisShenzhen doesn’t ease you in gently. The city hits you before you’ve found your bearings — buildings that make every other skyline feel like a rehearsal. This is the city that Tencent calls home, and the perfect backdrop for the Prosus Global Summit — where our Group CEO Fabricio Bloisi said what many of us needed to hear out loud: “The world has never seen so much change. We have to lead on that.” So we leaned in. We heard from economists, founders and operators building at a pace that reframes what’s possible. China’s AI agent user base: 600 million and climbing. Automotive development cycles slashed from 36 months to 18. R&D timelines from three months to three days. Not forecasts. Reality. We learned that behind the speed is a philosophy. John Fu put it simply — in Chinese culture, you preheat the wok before you cook. Relationships work the same way. Warmth first, then the work. We watched AI that doesn’t just assist. It builds. It iterates. It ships. The gap between knowing AI is changing everything and actually feeling it closed this week. Martin Lau and James Mitchell of Tencent delivered a masterclass in long-term thinking — the most durable advantage in the AI era isn’t the best model. It’s the lowest friction between idea and impact. And the summit didn’t just talk about building — it showed it. Day one opened with 12 finalist teams on the Shenzhen stage, each selected from over 1,000 entries across our ecosystem. At the AGM in August, Fabricio set the challenge — 10,000 AI agents across our ecosystem. Our CHRO Gustavo Vitti turned that ambition into a competition that captured the imagination of our entire ecosystem. Six months later, we didn’t reach 10,000. We reached 60,000 and still counting. By the time the cocktail glasses were raised that evening, we had our winners — across Growth, Efficiency, Individual Efficiency and Customer Experience. And our Ultimate AI Innovator: team ECHO from iFood. Silicon Valley awaits you. This is what investing in your people looks like when it compounds. A heartfelt thank you to Rajesh and Daniella on my team for their partnership in bringing the AI Talent competition to life, alongside Raiane who led the programme with such dedication. And through all of it, one question kept surfacing: are we moving fast enough? Fabricio believes the biggest company in the world is still to be built. That it should be us building it. 90+ ecosystem companies. 45,000+ people. 100+ countries. We already have everything we need to lead. The only question is whether we choose to. I leave Shenzhen energised, inspired and genuinely thrilled to be part of this journey. What a time to be doing this work. And to Gabriel and the entire Prosus team who brought this extraordinary event to life — thank you. You created something truly special. We felt it in every moment. Shenzhen showed us what choosing looks like. Now it’s our turn. #Prosus #Leadership #AI #China 🇨🇳 #Innovation #FutureOfWork #ProsusAITalent
-
Nihit Nirmal liked thisNihit Nirmal liked thisToday is an emotional day as I step down from the Groww board. Seeing the journey of the company from a 6 member team in Koramangala to a $13B publicly traded company over the last 7 yrs has been deeply fulfilling and I am grateful to the Groww founders (Lalit, Harsh, Neeraj, Ishan) for an incredible partnership. I am proud to step down with the company in such a strong position. At 88% YoY revenue growth and $300M annual net income run rate, it is witnessing an acceleration at scale! Groww provides both an inspiration and a peek into the future of Indian entrepreneurship. The markets in India are deepening and so are the ambitions and expertise of founders and management teams. I am excited for what the future holds for India's startup ecosystem and look forward to partnering with the next generation of founders.
-
Nihit Nirmal liked thisExcited to welcome Jillian Slyfield to Singapore and hear from her about the innovation and trends shaping the Tech landscape globally and in the US! Looking forward to hosting CHROs and Heads of rewards at this exclusive event where industry stalwarts Kulshaan Singh Puneet Swani will be speaking to CHROs and leaders around the impact of AI and therefore evolving role of HR. Rahul Chawla Ishita Goel Sara Tiew Sahil Batra Shikha GaurNihit Nirmal liked thisWe’re convening an exclusive group of senior HR leaders in-person on 12 May in Singapore for APAC Tech Community session focused on how AI, innovation and the CHRO agenda are evolving in practice. This closed, invitation only forum will bring together CHROs and senior HR leaders to exchange perspectives on: • AI and innovation trends from a US lens with implications for APAC • What it really takes to bring AI to life in organisations • The evolving role of the CHRO Peer and client case insights So excited for our upcoming event with Jillian Slyfield, Kulshaan Singh, Puneet Swani and industry leaders! Who should attend This session is designed for senior HR leaders in the technology sector, including: Heads of HR, CHROs / CPOs, Heads of Total Rewards, Global Heads of Compensation & Benefits. If you’re a senior HR leader in the technology industry and would like to hear more about how innovation is shaping the workforce agenda and how HR leaders are responding in practice, please reach out to Bhavana Chauhan, Ankur Saxena & Dhilip Kumar Raju. We’d be happy to share more details. Limited spots available. Share your interest today: https://aon.io/4t3tuxH 📅 Tuesday, 12 May 🕣 8:30 a.m. – 6:00 p.m., followed by networking drinks 📍 Singapore
-
Nihit Nirmal liked thisNihit Nirmal liked thisEvery few months, someone declares SEA venture "broken." Today, Salmon Group — a portfolio company we backed at Antler in an earlier round — announced a $100M financing: $60M in equity plus $40M of public bonds Significantly oversubscribed on the equity side, and placed cleanly on the debt side despite volatile global capital markets. One of the few genuinely oversubscribed rounds in the region this cycle. By coincidence, Tencent also announced a ~$500M investment into Kaspi.kz today — a Kazakhstan neobank - one of my favourite names in public markets, now sitting at a ~$17B market cap. In Kazakhstan. Population 20 million. The Philippines has 115 million. If a disciplined consumer fintech can compound to $17B there, the ceiling here is in a different class entirely — the underbanked opportunity is larger by every dimension. What Pavel, George, Raffy and the Salmon Group Ltd team have built in just three years is the real story: a BSP-licensed bank and a genuine challenger to legacy Philippine banks, with a capital structure more typical of a decade-old institution. Proud to have been in early. The next chapter should be the most interesting one yet. https://lnkd.in/gvcnQgxk Antler Martell Hardenberg Christopher Walsh Mathias Owing Maanum Keerat Singh Sethi
-
Nihit Nirmal liked thisNium recently opened our new office in San Francisco, and we were honored to have Jeffrey Siow, Acting Minister for Transport and Senior Minister of State for Finance for Singapore, join us to officially open the office. Moments like this make you pause and reflect on the journey. Nium was born in Singapore. Many of the opportunities we’ve had over the past decade, the ability to build, experiment, expand globally, and connect with the world, were made possible because of the incredible ecosystem Singapore has created for entrepreneurs. From early support and access to global markets to a regulatory environment that encourages innovation while maintaining strong standards, Singapore has played a huge role in shaping where Nium is today. As we continue to grow globally, it’s important to recognize where the journey started. A big thank you to Singapore and the many people who supported us along the way.Nihit Nirmal liked thisJust moved into a new space in San Francisco and what better way to mark the moment than welcoming friends from around the world into our home. We were honoured to have Singapore’s Acting Minister for Transport and Senior Minister of State for Finance, Jeffrey Siow, join us to officiate the opening. From Singapore 🇸🇬 to San Francisco 🇺🇸 , two global hubs we call home, it’s always special to open our doors and share these milestones together. Prajit Nanu, Andre Mancl, Anupam Pahuja, Rob Regan, Beto Lobos
-
Nihit Nirmal liked thisNihit Nirmal liked thisSometimes your wife pushes you to try things you would never attempt on your own. Last week she asked me to help with a small problem. Our kid is preparing for the IGCSE maths extended paper. The practice copy she had arranged came from a senior, and every question already had the answer written on it. She wanted a clean version so our kid could actually solve the paper herself. All I had were 20 pages of low quality phone screenshots. I opened Claude and asked it to rebuild the paper into a clean document. The text part worked well. The diagrams were the hard part. Claude first told me it could not generate the graphs and figures on its own so it has provided the text description of the images. My wife, who was acting as my client on this project, was not ready to accept the solution without the diagram images. That is where my engineering brain kicked in. I asked it to use Python libraries like matplotlib to draw every diagram. Circles, sectors, speed time graphs, Venn diagrams, cumulative frequency curves. All of it. A few hours later the full 20 page paper was sitting in front of me, exact diagrams and all. The side by side in the image tells the story. The real lesson for me was simple. Good AI use cases are sitting around us in everyday life. You just have to notice them.
Experience & Education
-
Yuuki
View Nihit’s full experience
See their title, tenure and more.
or
Publications
Projects
-
Micromax Logo Hunt
-
Crowd sourced activities to create new Micromax logo. We launched a global campaign for Micromax and drew in excess of 2500 entries from 1500 participants with minimal budget.
We did a mix of online and offline effort to drive such traction.
New Micromax logo was unveiled during a India vs Pakistan match during Asia Cup.Other creators -
Recommendations received
9 people have recommended Nihit
Join now to viewExplore more posts
Explore collaborative articles
We’re unlocking community knowledge in a new way. Experts add insights directly into each article, started with the help of AI.
Explore More