About
I'm the Co-founder and CTO of goodAds.ai, an AI-native creative automation platform…
Services
Activity
3K followers
-
Sakhil Chawla posted thisWe are hiring a Part-time Sales Account Executive at goodAds.ai. This role involves approximately 10–12 hours of outbound sales work each week, making it a great fit for those seeking part-time employment or a side hustle. Key Responsibilities: - Write and send outbound emails - Keep leads tracked and organized - Turn conversations into booked calls Ideal Candidate: - Proven experience in outbound sales - Organized and reliable - Based in North America (preferred) For more information, visit goodads.ai. If you believe you are a good fit, please send me a DM.
-
Sakhil Chawla posted thisHiring: Shopify Developer (3–5 yrs) — Remote, India 🇮🇳 We're hiring at GoodAds.ai, and I'm looking for a Shopify developer who genuinely lives in the platform — not someone who's brushed up against it, but someone who knows why things break and builds so they don't. Here's what "knows Shopify inside out" means to me: → Fluent in Online Store 2.0 — sections, JSON templates, metafields/metaobjects, Theme Blocks — and you've kept pace as Shopify moved toward Horizon. → Real experience with the GraphQL Admin + Storefront APIs (2026-04), webhooks, and app extensions. Not stitched-together snippets. → You understand Checkout Extensibility — Checkout UI Extensions, Web Pixels, Shopify Functions. You know Scripts are gone and checkout.liquid is on its last legs, and you know why that shift happened. → Bonus: you've shipped headless with Hydrogen/Oxygen, or built a custom/embedded app. → Double bonus: you're watching where Shopify is heading — agentic commerce, Storefront MCP, UCP — not just where it's been. But past the checklist, the thing I actually care about: you treat a merchant's store like it's your own revenue on the line. Careful, fast, and allergic to fragile hacks. 3–5 years of real Shopify work. Remote, India-based. Competitive comp. If this is you → DM me, and show me something you've shipped. If you know someone who fits → tag them or send them my way. A good referral genuinely means a lot to me. 🙏
-
Sakhil Chawla shared thisWe're hiring a Customer Success / Performance Manager at Socioh 🚀 If you're passionate about eCommerce advertising, know your way around Meta ads, and thrive in a fast-moving startup — this one's for you. What you'll be doing: → Owning client relationships end-to-end → Coordinating across design, tech, and sales teams → Monitoring ad accounts and solving problems proactively Bonus points for Shopify, Meta Business Manager, or ad agency experience. Full-time, remote. US hours required. Drop a comment or DM me directly with your experience. Would love to connect. I’m #hiring. Know anyone who might be interested?
-
Sakhil Chawla shared thisMost leaders are running monolith organizations in an agentic world. Everything tangled. One decision blocks five others. One person leaving collapses a domain. The leaders winning in 2026 think about their org differently — as a multi-agent operating system. You = the context engineer, not the bottleneck. Route the right signals to the right domains. Then get out of the way. Domains = autonomous agents, not departments. Sales. Product. Eng. People. Finance. Legal. Each runs independently with clean interfaces between them — no silos, no single points of failure. Playbooks = your agentic workflows. What lets domains operate without you? Repeatable, documented plays per function. That's it. Feedback loops = your compound advantage. Agentic systems win because they close loops fast. The best orgs win the same way. Single-agent thinking scales linearly. Multi-agent org design scales exponentially. This is what AI-native leadership looks like. Are you orchestrating or still executing? #AgenticAI #ContextEngineering #MultiAgentSystems #AILeadership #FutureOfWork #AIOrchestration #AITransformation #Leadership
-
Sakhil Chawla shared thisWhat if every developer on your team could learn from the best engineer's workflow — in real time, without a single meeting? That's what happens when your tooling lives in the open. We built our internal marketplace repo on exactly this principle: → Every command, query, and automation runs visibly — in shared channels, not private windows → Patterns that work spread naturally. Good technique becomes team muscle memory → New team members don't wait for onboarding docs — they scroll back and see how work actually gets done → The repo gets smarter over time because everyone who uses it improves it The result isn't just a faster workflow. It's a living knowledge system. Junior developers absorb senior judgment by proximity, not by asking for it. Teams stop reinventing the same solutions. The gap between "how we say we work" and "how we actually work" disappears. There's a concept in German craft culture called Lehrwerkstatt(Referred from Tobias Lütke River post) — a teaching workshop where the shop floor is the classroom. You learn by being near the work. We think the best engineering orgs are built the same way. Not through training programs. Through radical visibility. What does your team's shop floor look like? #engineeringculture #developerexperience #softwaredevelopment #teamlearning #buildinpublic #goodAdsAI
-
Sakhil Chawla shared thisWe're hiring engineers at goodAds.ai(India location). We're not hiring "AI Engineers." Every VC deck and Twitter thread has been pushing the title with a lot of fanfare. I agree with the general insight. Andrej Karpathy nailed the big idea: "the hottest new programming language is English." Every engineer's interface to the machine just changed. This resonates 100%. But I'm not latching onto the "AI Engineer" title (yet) for two reasons. 1.) Every engineer should be AI-native first. Our whole team cranks with Claude Code, Cursor, and the Anthropic SDK. Backend, frontend, infra — all of them. That's the precondition for a dedicated AI builder to actually move the needle. Specialize too early and you've just hired one person to do what everyone else now politely refuses to learn. 2.) "AI Engineer" is silo-creating. The pre-LLM org chart didn't serve product velocity. Putting one "AI Engineer" next to a backend engineer still pasting tracebacks into Google recreates the exact problem we're trying to escape. If everyone has the capability, the title is just a flag for who gets paged when prod hallucinates. The data backs broad fluency over titles. Shopify made AI usage a baseline performance expectation. JPMorgan rolled an internal LLM assistant out to 60k+ employees. Stripe ships Devin into engineers' workflows by default. For the 50 fastest-shipping startups I track, "engineers using AI tools daily" outnumber "AI Engineer" job postings by 30x+. The title will eventually emerge for the deep-systems work — agent infrastructure, eval harnesses, the unglamorous plumbing of making models reliable. Fine. But that's a specialization, not a starter hire. If you treat Claude / Cursor / the Anthropic SDK like a power tool — not a novelty — and you want to ship at the pace this stuff actually unlocks, we should talk. DM me, or drop your GitHub in the comments.
-
Sakhil Chawla shared thisLast week I deleted 9,170 lines of agent skills from one of our repos. We didn't lose any capability. We added a few. This was the most leveraged hour of AI tooling work I've done all quarter, & almost none of it was code. The setup: most teams adopting agent-driven dev end up with skills in three places — every engineer's `~/.claude/skills/`, the repo's `.claude/skills/`, & a Confluence page someone started in February. Within a quarter, the three diverge. The agent picks up a stale local copy. PR reviews go sideways. Nobody can answer which version is canonical because there isn't one. We fixed it by drawing one hard line: the marketplace is canonical, the repo holds nothing but the data the marketplace skills read. Internally we call this Pattern B. Concretely: ��� 9 plugins live in a private marketplace repo. `goodads-frontend`, `goodads-backend`, `goodads-core`, `goodads-testing`, `goodads-safety`, plus four more. → Skill bodies (the actual SKILL.md files) live in the marketplace, nowhere else. → The repo keeps only support files — a review checklist, an issue taxonomy, a design-token mapping. Data, not behavior. → The marketplace skill reads the repo file first, falls back to its bundled default if absent. Skills work in any repo; rich repos just give them better inputs. → CI on the marketplace runs three gates on every PR — a skill linter, an AgentShield safety scan, a human review. What this unlocks: 1. **One source of truth.** When the design system updates, `create-component` updates once. Engineers pull, agent gets the new behavior. 2. **Safety gating.** AgentShield catches prompt-injection patterns, secrets accidentally pasted into CLAUDE.md, overly broad tool allowlists. Nothing ships to engineers without passing. 3. **Discoverability.** Engineers find a skill by browsing the marketplace, not by guessing which teammate wrote one for X. 4. **Deprecation.** Friday Agent Governance Board kills any skill with <3 invocations in a quarter. Without a marketplace there's no surface to deprecate *from* — dead skills accumulate forever. The trade-off I won't pretend doesn't exist: indirection. When a marketplace skill reads a file from the repo, an engineer debugging behavior has to look in two places. We mitigate it by listing the support-file path in the skill's description & in the repo's root CLAUDE.md. It's still a real cost. But it's a cost I'd pay a hundred times before going back to "every engineer maintains their own dotfiles." . If you're past 5 engineers & your skills aren't centralized yet — you don't have a tooling problem yet. You will, in about two months. The npm registry analogue lands here: public packages for breadth, private registry for the things that make your team your team. Skills are the same shape. The teams that figure that out before the painful quarter compound their internal knowledge faster than anyone else. What pattern is your team on? #AIEngineering #DeveloperTools #AgenticAI #StartupCTO #ClaudeCode
-
Sakhil Chawla shared thisEveryone told us to fine-tune. We were building an AI creative system for hundreds of brands across every major ad platform — each brand with a different visual identity, each platform with different constraints. The “obvious” play was per-brand fine-tunes. We didn’t. We launched on frontier models with a structured intermediate representation that constrains output before anything visual gets generated. The reasoning: → Fine-tuning is expensive to iterate. Brands change their look every season. Retraining loops kill velocity. → Frontier models compound. Every release makes our system better for free. Fine-tunes don’t — they’re a snapshot. → The bottleneck isn’t model capability. It’s representing brand identity in a structured way the model can follow. Will we fine-tune later? Probably, for narrow tasks where frontier inference is overkill. Not as the core engine. The lesson I keep relearning: when models are improving this fast, the right architecture is the one that lets you ride the curve, not the one that locks you off it. Every quarter since launch, this bet has looked smarter. Opus 4.7 dropped, our outputs got better. We did not retrain. We did not redeploy. The structured layer absorbed the upgrade. The architecture you can throw away is more valuable than the architecture you optimize. #AI #AIEngineering #FoundationModels #StartupCTO #MachineLearning
-
Sakhil Chawla shared thisEvery Friday at goodAds, for 15 minutes, we look at one table. One column: the skill name. One column: invocations in the last 90 days. Anything below 3 goes on the kill list. No defense. No "but I used it that one time on the Tuesday thing." Below 3, you're out. This is the single most important governance mechanism we added to our Claude Code marketplace — more than the cost controls, more than the safety hooks, more than the MCP data layer. Because the failure mode of AI tooling in 2026 isn't "AI doesn't work." It's **"we accumulated 136 skills, 94 of them are unused, and the model has to read all of them to pick one."** SuperClaude shipped 136 skills and the community quietly stopped using most of them. We are not going to ship 136. We are going to ship 20 good ones per plugin and delete the rest every Friday. How we run it: → Auto-scored via `/eval plugins/` — PluginEval from wshobson gives each skill a judge score. Anything under 0.5 also joins the kill list. → Two-week grace window. If a killed skill had a champion, they have until the next Friday's Friday (i.e. two Fridays later) to make a case. Otherwise, gone. → Output is a single markdown decision log at `docs/retros/YYYY-MM-DD-friday-retro.md`. Version-controlled. Searchable. Auditable. → The retro itself is a `/friday-retro` slash command in our `goodads-core` plugin. The governance is the tooling. Your AI tooling gets worse every week you don't prune it. Prune it. — CTOs / founders trying to keep their AI tooling from becoming an accumulating graveyard — reach out. I'll send you our `/friday-retro` skill and the decision template. #AIEngineering #ClaudeCode #EngineeringMetrics #CTO #ToolingStrategy
Sakhil Chawla replied to a comment
3mo
Alyssa Milot thankyou
-
Sakhil Chawla liked thisSakhil Chawla liked this16 years later, it felt wonderful to be back at Thapar Institute of Engineering & Technology alumni meet! The Delhi Regional Alumni Meet was an evening of reconnecting and exchanging experiences. Some particularly enriching conversations were with a professor from IIT Delhi, the South Asia Leader at AMD, and the former Operations Head of the Mohali semiconductor plant. Suprisingly, several Thaparians from the 1970s batches had themselves been associated with the semiconductor industry. A striking reminder that while India is now working to build its semiconductor capabilities and catch up with the US and China, our alumni have been part of this journey for decades. A legacy our current students can learn from. Also great to reconnect with my batchmate Puneet Jindal , who is doing pioneering work in AI and deep learning through Labellerr AI , a startup incubated at Thapar Innovate An evening of learning from collective experiences and drawing inspiration for the future. Thanks to Mansi Bhargava and the Thapar Alumni Relations Office team for bringing the community together. Rajeev Ranjan Vederah Padmakumar Nair Ajay Batish #ThaparAlumni #Thaparians #Semiconductors #AI #DeepLearning
-
Sakhil Chawla liked thisSakhil Chawla liked this🚿 Sauce leads Rs 15 Cr funding round in kitchen and bathroom brand Lorazzo Lorazzo, a design-led kitchen and bathroom brand, has raised Rs 15 crore led by Sauce, with Sprout Venture Partners, Panthera Peak and Alteria Capital also participating. 📌 Funds will go toward product development, team building and expanding its online and offline presence across more states. 🏢 Co-founded in 2024 by Jatin Luthra and Saurabh Gupta, Lorazzo makes faucets, sinks, shower systems, bidets and smart toilets, its Intelli Smart Commode is priced at Rs 58,999, versus Rs 1.5-5 lakh for comparable international brands. 📈 The brand has grown revenue 6X in 12 months and is targeting Rs 100 crore in revenue over the next two years. Full story: https://lnkd.in/gJ4wwzvZ #Funding #Lorazzo #D2C #Entrackr
-
Sakhil Chawla liked thisSakhil Chawla liked thisFrom #comedy and #content creation to becoming one of India’s most interesting #angelinvestors. Across the companies mapped, Tanmay has backed #24 startups spanning #AI, #fintech, #creatortools, #consumer, #climatetech and #Web3. 1) Health and climate, Ultrahuman and Alt Carbon 2) Fintech and wealth, Wint Wealth, EquityList, Trackk, @infinyteclub and @Timeswap 3) AI and software, Superleap AI CRM, GrowthSchool and Unikon.ai 4) Creator economy, Rigi , @crowdpad , SuperShare and qoohoo 5) Consumer, fashion, community and Web3 mainstreet., Iztri, Flint , @Mean DAO , Samudai.in, AvalonMeta Ventures, @Offline, HYPD and rep3. A traditional investor may start with: Market size → business model → financial projections A creator-turned-investor can sometimes start somewhere different: Is this behaviour actually changing? Will people care about this product? Can this company build a community around itself? Can the idea spread? That doesn't replace traditional venture analysis. But it can add a very different lens. And perhaps that is why we are seeing more operators, founders and creators becoming active angel investors in India's startup ecosystem. They are not only providing capital. They bring distribution, audience, brand understanding, founder empathy and access to networks. #TanmayBhat #AngelInvesting #VentureCapital #Startups #IndiaStartups #Fintech #ArtificialIntelligence #CreatorEconomy #ClimateTech #Web3 #XponentialVentures
-
Sakhil Chawla liked thisSakhil Chawla liked this𝗖𝗘𝗟𝗘𝗕𝗥𝗔𝗧𝗜𝗡𝗚 𝗔 𝗧𝗛𝗔𝗣𝗔𝗥𝗜𝗔𝗡’𝗦 𝗡𝗘𝗪 𝗖𝗛𝗔𝗣𝗧𝗘𝗥 🌍 We are delighted to celebrate our alumnus and Institute Ambassador, Arijit Chatterjee, as he begins an exciting new chapter with 𝗦𝗻𝗼𝘄𝗳𝗹𝗮𝗸𝗲, leading 𝗘𝗺𝗲𝗿𝗴𝗶𝗻𝗴 amp; 𝗚𝗿𝗼𝘄𝘁𝗵 𝗦𝗜 𝗣𝗮𝗿𝘁𝗻𝗲𝗿𝘀𝗵𝗶𝗽𝘀 𝗶𝗻 India.After six impactful years at AWS, Arijit now moves to the intersection of 𝗗𝗮𝘁𝗮, 𝗔𝗜 𝗮𝗻𝗱 𝗣𝗮𝗿𝘁𝗻𝗲𝗿𝘀𝗵𝗶𝗽𝘀, helping build an ecosystem that can translate technology into meaningful outcomes for businesses and customers.For us at Thapar, there is an additional reason to celebrate. Arijit has continued to remain deeply connected with his alma mater, returning to campus, engaging with our students and entrepreneurial ecosystem, and giving back through his experience and leadership. 𝗦𝘂𝗰𝗰𝗲𝘀𝘀 𝗶𝘀 𝗮𝗹𝘄𝗮𝘆𝘀 𝘄𝗼𝗿𝘁𝗵 𝗰𝗲𝗹𝗲𝗯𝗿𝗮𝘁𝗶𝗻𝗴. 𝗦𝘂𝗰𝗰𝗲𝘀𝘀 𝘁𝗵𝗮𝘁 𝗰𝗼𝗻𝘁𝗶𝗻𝘂𝗲𝘀 𝘁𝗼 𝗴𝗶𝘃𝗲 𝗯𝗮𝗰𝗸 𝘁𝗼 𝗶𝘁𝘀 𝗿𝗼𝗼𝘁𝘀 𝗶𝘀 𝗲𝘃𝗲𝗻 𝗺𝗼𝗿𝗲 special.Congratulations, Arijit, on this wonderful new beginning. We wish you tremendous success as you build, lead and create impact in this next chapter. 𝗢𝗻𝗰𝗲 𝗮 𝗧𝗵𝗮𝗽𝗮𝗿𝗶𝗮𝗻, 𝗮𝗹𝘄𝗮𝘆𝘀 𝗮 𝗧𝗵𝗮𝗽𝗮𝗿𝗶𝗮𝗻. ❤️ #ThaparInstitute #TIET #ThaparAlumni #GlobalThaparians #ArijitChatterjee #Snowflake #DataAI #ArtificialIntelligence #Partnerships #AlumniPride
-
Sakhil Chawla liked thisSakhil Chawla liked this𝗖𝘂𝗹𝘁𝘂𝗿𝗲 𝗶𝘀 𝗻𝗼𝘁 𝗰𝗿𝗲𝗮𝘁𝗲𝗱 𝗯𝘆 𝗽𝗼𝗹𝗶𝗰𝗶𝗲𝘀 𝗼𝗻 𝗽𝗮𝗽𝗲𝗿. 𝗜𝘁 𝗶𝘀 𝗰𝗿𝗲𝗮𝘁𝗲𝗱 𝗯𝘆 𝗽𝗲𝗼𝗽𝗹𝗲, 𝗶𝗻 𝗺𝗼𝗺𝗲𝗻𝘁𝘀 𝘁𝗵𝗮𝘁 𝗯𝗿𝗶𝗻𝗴 𝗲𝘃𝗲𝗿𝘆𝗼𝗻𝗲 𝘁𝗼𝗴𝗲𝘁𝗵𝗲𝗿. 🏏 Today, the 𝗩𝗲𝗱𝗮𝗻𝘁𝘂 𝗢𝗻𝗹𝗶𝗻𝗲 & 𝗢𝗳𝗳𝗹𝗶𝗻𝗲 𝘁𝗲𝗮𝗺𝘀 came together for a cricket match, with our founders and leadership playing alongside the team. Just 𝗼𝗻𝗲 𝗺𝗼𝗻𝘁𝗵 𝗶𝗻𝘁𝗼 𝗩𝗲𝗱𝗮𝗻𝘁𝘂, and this is one of those moments that stays with you. Because when leadership stands on the same ground, plays the same game and shares the same laughter, 𝗵𝗶𝗲𝗿𝗮𝗿𝗰𝗵𝘆 𝘁𝗮𝗸𝗲𝘀 𝗮 𝗯𝗮𝗰𝗸 𝘀𝗲𝗮𝘁 𝗮𝗻𝗱 𝗽𝗲𝗼𝗽𝗹𝗲 𝗰𝗼𝗺𝗲 𝗳𝗶𝗿𝘀𝘁. Great organisations are not built only in boardrooms. They are built when 𝗽𝗲𝗼𝗽𝗹𝗲 𝗳𝗲𝗲𝗹 𝗰𝗼𝗻𝗻𝗲𝗰𝘁𝗲𝗱, 𝘃𝗮𝗹𝘂𝗲𝗱 𝗮𝗻𝗱 𝗽𝗮𝗿𝘁 𝗼𝗳 𝘀𝗼𝗺𝗲𝘁𝗵𝗶𝗻𝗴 𝗯𝗶𝗴𝗴𝗲𝗿 than their individual roles. 𝗪𝗼𝗿𝗸 𝘁𝗼𝗴𝗲𝘁𝗵𝗲𝗿. 𝗣𝗹𝗮𝘆 𝘁𝗼𝗴𝗲𝘁𝗵𝗲𝗿. 𝗚𝗿𝗼𝘄 𝘁𝗼𝗴𝗲𝘁𝗵𝗲𝗿. 🏏 Pulkit Jain Sir Sahil Bhatia Sir Sridhar R Sir Deenbandhu Pandey Sir Vedantu Abhinav Srivastava Sir Arnab Dutta Sir #Vedantu #Culture #Leadership #Teamwork #WorkplaceCulture #OneTeam
-
Sakhil Chawla liked thisSakhil Chawla liked thisHibbett just showed what it looks like when AI actually earns a shopper's trust. The team launched Agentic Commerce Search (fka Cimulate) and followed it with Hibbett AI — a full conversational Shopper Agent. And shoppers didn't just try it, they leaned on it: 90% of Hibbett's core shopper journeys are now handled by Hibbett AI. A few things that stood out to me: - Shoppers can ask for exactly what they want — like "running shoes for wide feet” — and get real, relevant recommendations back - Discovery and action live in one experience, so browsing turns into buying without extra friction - A seamless handoff to customer service for anything outside shopping, so the agent knows its lane Congrats to the entire Hibbett team — check out what they built at hibbett.com. Hibbett
-
Sakhil Chawla liked thisExcited for this one and proud of the team for delivering it. Lots more cool “Smart Window” stuff in the pipe we can’t wait to ship too.Sakhil Chawla liked thisSome partnerships can break assumptions. Today Mozilla and Mistral announced an exciting partnership built around the simple bet that great AI models should be able to work across products, and great products should be able to work with different models. This is how the internet should be trending. Firefox doesn’t need to build AI models. Mistral doesn’t need to own the browser. Instead, we can work together to build something useful and give people real choice. Big AI doesn’t need to dictate all our choices when open-source, independent, responsible AI and browser companies collaborate to compete and win. The browser is becoming one of the most important AI distribution points. I’d rather build choice into the Firefox user experience now and set the bar for what’s possible rather than wait for it to be mandated. It’s a good day to be building Firefox. It’s a good day for the open web. I’ll let The Wall Street Journal tell the rest of the story. Link below.
-
Sakhil Chawla liked thisSakhil Chawla liked thisLast Saturday I spoke at Product Tank event on From Feature Owner to Ecosystem Enabler — the shift I've lived through as a PM, and one I think most of us hit eventually. The core idea: product thinking solves a problem; platform thinking enables many solutions. One builds a road. The other builds the grid that lets a thousand journeys happen you never planned for. That change shows up as four shifts, each one leading to the next: → From user needs to ecosystem tensions — you're now serving parties whose interests compete. → From features to compounding capability — as those tensions repeat, you build primitives instead of one-offs. → From speed to trust — to let others build on that capability, trust becomes the real constraint. → From requirements to incentives — since you can't mandate behaviour across an ecosystem, you design the payoff instead. Which leads to the question I left the room with: the point isn't "Am I on a platform team?" — it's "Am I building something others depend on to create value?" If the answer is yes, the platform mindset already applies, title or not. Thank you, ProductTank Bengaluru, for a genuinely engaged room and sharp questions. #ProductManagement #PlatformThinking #ProductLeadership #Ecosystem
Experience & Education
Licenses & Certifications
Honors & Awards
-
Winner - SAP CD Global Innovation Challenge
SAP CD
1 ) Conceptualized, designed and implemented a prototype for SAP – CD global innovation challenge over a period of 4 months for a SAP - Retail solution which SAP believes that it can help SAP reach its target of 1 billion users by 2015
2) The solution comprised of a web application and an android app as part of prototype.
3) Won the innovation challenge final in Germany and the company sponsored the project for converting the prototype into final product.
Recommendations received
5 people have recommended Sakhil
Join now to viewOther similar profiles
Explore 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