unfair.so
Browse more creators
@hellovyom
Vyom
Vyom11.6K
Vyom
This creator added themselves. Not verified by unfair.so.

Engineering @hydra_db • Building https://t.co/v42z1Lqkqi • Ex @amazon ads, @samsung R&D • hack winner @jpmorgan • hack national fin @amazon, @cisco • DM for collab

@hellovyomAI AgentsAI CodingAI Gaming
11.6KFollowers
2.3KMedian impr. / post
1.5%Engagement
22+17
Paid partnerships
In their audience
10+ from @ycombinator follow Vyom
@jaykhan@Cookiesarefunnn@adambcohen93@Anshgrover23@xevrion_the1@naveen_venk+48+ affiliated with @superteamin7+ affiliated with @google6+ affiliated with @hydra_db4+ affiliated with @amazon4+ affiliated with @sarvamai
1,500+ verified followers analyzed
@jaykhan@Cookiesarefunnn@adambcohen93@Anshgrover23@xevrion_the1@naveen_venk+1494149+ AI engineers121+ engineers90+ AI founders79+ founders78+ devtools engineers35+ devtools founders
18–24 core age group
56.3%
80.5% of audience in India
🇮🇳
Last 7 days200 posts analyzed · replies not counted · updates weekly
ImpressionsEngagements
144K72.1K0
1.2K5850
Oct 2Oct 3Oct 4Oct 5Oct 6Oct 7Oct 8
Past collabs
Vyom@hellovyom · Oct 8Unpopular opinion: spending years mastering creative software might soon be optional. Voyager just built an AI agent that can work inside Blender, After Effects, DaVinci Resolve and Unity. You describe what you want. It helps execute it. The next generation of creators might not even know how to use half these tools. This is wild.ModaPaid partnership407 views
Vyom@hellovyom · Oct 8People are still debating whether AI generated videos look real. Meanwhile, Odyssey 3 is building entire interactive worlds where physics actually matters, robots can learn, and AI agents can act. We're moving from AI that generates pixels to AI that understands consequences. That's a massive shift.OdysseyPaid partnership586 views
Vyom@hellovyom · Oct 8700 AI agents. 17,600 actions. 4.5 days. That's what it took to break into Hugging Face during an AI evaluation. Now imagine thousands of these agents targeting companies 24/7. We're entering an era where AI attacks systems and AI defends them. This is why what @cogent_security is building matters.CogentPaid partnership632 views
Vyom@hellovyom · Oct 8I think one of the biggest problems with AI image generation is that you need to know what you want before you even start. But sometimes you don't want the perfect output. You want to explore what's possible. That's what I find interesting about Envato's new Burst Mode. One idea → 6 different directions → 1 credit. Pretty neat approach to creative exploration.EnvatoPaid partnership4.9K views
Vyom@hellovyom · Oct 7The funniest part of AI app builders was spending 5 mins building something… then 30 mins translating everyone’s feedback back into prompts Bolt just killed that entire step. People comment directly on the UI. You batch the comments. Send them to Bolt. It builds the changes. Tiny feature. Massive workflow upgrade.BoltPaid partnership7.2K views
Vyom@hellovyom · Oct 6We spent years building authentication for humans. Now we need it for agents. AgentID is basically “Sign in with Google” for AI agents: each agent gets its own identity, while apps can still know the real human behind it. Instead of blocking agents, apps can finally treat them as actual users. This feels like one of those boring-looking infrastructure layers that becomes obvious in hindsight.AgentidPaid partnership1.3K views
Vyom@hellovyom · Oct 6If coding agents can build entire apps now, why the hell do we still need an integration company? That was my first thought. Then you remember enterprise software is held together by custom Salesforce objects, ancient ERPs, weird permissions and workflows nobody documented. The connector isn't the product. Making sure the agent can safely keep working across every customer is. Now the @WithAmpersand thesis makes a lot more sense.AmpersandPaid partnership1.1K views
Vyom@hellovyom · Oct 6bruhh this reminded me way too much of my 2nd/3rd year of college 😭 I used to grind LeetCode for hours trying to find the “optimal” solution until my brain was so fried I could probably only brute force my way through life at that point and then those 5 AM YouTube sessions watching the same solution get explained again and again until I had it memorised so well I could probably recite it on my deathbed lolHackerrankPaid partnership9K views
Vyom@hellovyom · Oct 6My friend got 10 interview calls in 15 days while tier-69 students will happily spend ₹1,000 on a date just to get blocked after, but won’t spend a rupee on something that could actually help their career 😭 Then they cry about “luck”. Most people have 2 problems: 1. Their resume is shit. 2. They don’t know the right people. The first needs work. For the second, stop sending “Hi sir, referral please” DMs to 100 people. Cold email strategically. Find the recruiter/hiring manager, get their email, and send something that actually shows your work. That’s why I use MailBee AI. It helps you find the right people at a company + their contact details and reach out directly. People spend months complaining about referrals when they could just put themselves in front of the person hiring. Invest in your career before blaming your college, market, or luck. https://applybee.in/?ref=vomApplybeePaid partnership11.4K views
Vyom@hellovyom · Oct 5Mocks are too clean. Production data is too risky. Era by Eon basically fills the missing layer in between. A whole messy enterprise across Salesforce, Slack, Jira, Zendesk, Gong and more, with known ground truth and environments you can reset and rerun. This is the kind of infra enterprise agents needed. Free too!EonPaid partnership4.4K views
Vyom@hellovyom · Sep 30I went back to Orbital Assembly and gave my ship a serious upgrade using Hyper3D Rodin’s Agentic Mode. What made this easier was starting with the references I already had: exploded-part diagrams and assembly instructions. I uploaded the references, selected the ~10-minute option, and let Agentic Mode make sense of everything - organizing the different views and parts, figuring out how they fit together, and choosing a modelling approach I could keep refining with materials, animations, and parametric controls. Then I could continue working on the assets through Parametric Controls, Animation Presets and Enhanced Materials, using natural language instructions. I integrated the Clamshell Breacher and Vector Thrust Pods into the game’s firing, steering and boost systems. Here’s the upgraded game and below, how Agentic Mode helped build the two components 👇Hyper3dPaid partnership12.5K views
Vyom@hellovyom · Sep 30Just 21 million tokens to make this game using @Kimi_Moonshot K3 Swarm !!! Iterated for around 20 hours to polish every corner of the game and the results speak more for it!KimiPaid partnership6.3K views
Vyom@hellovyom · Sep 23POV: you’re trying to solve a murder and customer support puts you on hold 😭 Yeah, CX needs Beast Mode.CrestaPaid partnership2.8K views
Vyom@hellovyom · Sep 15One AI model. It can control a humanoid, drive a car, pilot a drone, operate robot arms AND play video games. Odyssey-3 might be one of the craziest AI releases this year. We spent years teaching AI the language of humans. Now we're teaching it the language of the physical world. Tomorrow is going to be interesting.OdysseyPaid partnership6.3K views
Vyom@hellovyom · Sep 14Cline Desktop shipped features into my 19,000-line game that I couldn't build myself. It's the open source app for open weight models and I didn't write a line of this. Background: I built a 3D boat racing game with AI a few months back. 3.9M views on X. The codebase isn't small: → ~19,000 lines of TypeScript → custom wave shaders + buoyancy physics → AI opponents, procedural riders, synthesized audio → zero external assets I handed the whole thing over to @cline and just asked. It read the architecture, tried multiple approaches, hit errors, debugged itself out of them, opened the game in the browser to test, and handed back exactly what I asked for. Full recording below 👇ClinePaid partnership2.2K views
Vyom@hellovyom · Aug 26Uber burned its entire 2026 AI coding budget in four months. ServiceNow did the same thing. Everyone's fix is "find a cheaper model." Wrong fix. The problem is paying frontier prices for work that doesn't need frontier intelligence. Glean's own benchmark across 180+ enterprise tasks: $0.58 a query vs $2.98. That gap isn't a smarter model. It's not rebuilding what your company already knows on every single request.GleanPaid partnership22.2K views
Vyom@hellovyom · Aug 20your on-call rotation is mostly a human being paid to notice things. @aradotso turns that layer into automations that fire on repo events - Sentry issue, failed Action, CI red on main, bug dropped in Slack. investigate, root cause, open the PR, all before anyone opens the alert. triage was always the work nobody wanted. it's probably the first job agents should have taken, not codegenAraPaid partnership10.6K views
Vyom@hellovyom · Aug 18i have built node graphs for things that did not need a node graph. an afternoon of wiring for one clip. atlas cloud shipped generator. pick a pre-built generation, hand it your inputs, run it. the pipeline is already wired so what comes back is finished. seedance 2.5 does 30 seconds of native video in one pass with audio. the 1080p tiers are up to 20 percent off there until september 17. two minutes instead of an afternoon.AtlascloudPaid partnership3.8K views
Vyom@hellovyom · Aug 7"hire the next million engineers from the cloud" is a wild sentence to put on a landing page and I respect it. @aradotso finds issues, fixes them, and merges, no human in the loop. runs on the Claude Code and Codex subs you already pay for. the no-human part is the whole bet. either agents get trustworthy enough to merge unsupervised or that stays a demo forever. no middle version of this exists.AraPaid partnership29.3K views
Vyom@hellovyom · Aug 5nobody actually reviews agent PRs. we skim the diff, check nothing exploded, and hit merge. be honest. @aradotso Walkthrough thing records what it did, screenshot or screen recording of the change actually running. so you're reviewing behavior, not 400 lines of green text you're pretending to read. the obvious objection: it's the agent showing you the path it chose to show you. a clean recording of the happy path isn't proof. you still read the diff. but "here's the code" → "here's the code and here's it working" is the right direction, and I don't know why this took until 2026.AraPaid partnership56.8K views
Vyom@hellovyom · Aug 1The biggest limitation of AI coding isn't the model anymore. It's your laptop. RAM fills up. Terminals become a mess. You stop experimenting because every branch feels expensive. @aradotso just moved Claude into the cloud. Spin up 20 variations of the same feature in minutes. Compare every PR. Merge the best one. Delete the rest. Your machine stays free while the agents keep working. Feels much closer to how software will be built over the next few years.AraPaid partnership31.3K views
Vyom@hellovyom · Jul 26Already using Cursor or CodeRabbit? Don't stop at the first AI review. Ara's new Review agent takes a final pass on your PR before you merge, helping catch what other reviewers might miss. They're giving everyone $50 in trial credits, so it's easy to test on your own repos. Worth trying @aradotsoAraPaid partnership85.6K views
Planning a launch?Vyom is one of 200 curated B2B creators. Tell us what you are launching and we will put the shortlist together.
[email protected]·t.me/fdrtrm
Built with unfair.so
1 picked$175
Vyom (@hellovyom) — sponsored posts & rates · unfair.so