ai agents & integrations
add agents, smart model switching, or knowledge retrieval to your existing product. i contribute to mastra (ai agent framework) and use vercel ai sdk in production daily.
i help startups add ai to their products
agents, integrations, or full ai-powered mvps. i contribute to the typescript tools i use — so when things break, i can actually fix them.
most founders spend their first month watching demos break because the llm hallucinated, debugging tool calls that worked yesterday, and rewriting code for frameworks that changed last week. i've already shipped through all of that.
the deal
$500for 20 hours of work
that's your ai integration or mvp kickstart. 20 hours gets you a working ai chat feature, an agent integration, or a solid prototype. not satisfied? full refund — i'd rather lose the money than have an unhappy client.
production-ready ai features — error handling, retries, and monitoring included
clean typescript codebase your next developer can extend without rewriting
documentation so you can maintain it yourself or onboard teammates
what i can help with
ai-powered mvps
build your ai product from scratch. chatbots, automation tools, ai-native apps. production-ready with proper error handling and observability.
tool adoption
migrate to modern ai frameworks before your competitors do. you get production code while others are still watching tutorials.
prototype cleanup
upgrade your cursor/bolt prototype into something production-ready. add tests, fix the architecture, make it reliable before launch.
what you actually get
1
a framework, not a dependency
you'll stop being dependent on contractors for every ai feature. i'll teach you how to build on your own — that's freedom, not a subscription.
2
systems that run themselves
i'll set up agents and automation that handle repetitive tasks. your product should work while you sleep.
3
honest assessment
if something is out of my depth, i'll tell you. if your idea has technical problems, i'll tell you that too.
what i expect in return
the price is low intentionally. in exchange, i have three requirements:
testimonial
public feedback on x or linkedin. positive or negative — doesn't matter. i want honest reviews.
brutal feedback
tell me what sucked. tell me what worked. i need to know where i'm weak to get better.
one introduction
if you liked my work, introduce me to one founder friend who's adding ai to their product.
why work with me now
i'm a software engineer with 3.5 years building backend systems at fintech and saas startups. i'm building my consulting practice from scratch — which means you get senior-level ai expertise at early-stage prices.
what you're getting: when your ai feature hits a weird bug, i can read the framework source code and fix it — not just google the error message. i'm a mastra contributor, so i know how it works under the hood. the mastra team refers clients to me when they need implementation help.
my edge: i shipped production mastra code within 2 months of the framework launching. i adopt tools before they're mainstream — you get working implementations while your competitors are still figuring out the basics.
proof of work
built an ai agent for this yc-backed startup's real-time diagram editor. natural language prompts to canvas mutations, with tool calls, safety checks, and audit logs. shipped to production — now handling customer requests daily.
10 merged prs focused on memory systems and integrations. when you hit a weird bug or edge case, i can read the source code and fix it — not just google the error message.
how i got my first client
i didn't cold email or run ads. the mastra cto saw my contributions and connected me directly.
that referral turned into artifact.engineer — my first consulting project. this is what i mean when i say the mastra team trusts my work.
technical stack
this stack means your ai features plug directly into your next.js or react app — no awkward glue code, no performance bottlenecks, no "it works on my machine" surprises.
ai tools:mastra, vercel ai sdk, openai apis, langchain, rag pipelines
core stack:typescript, node.js, postgres, next.js, react
infra:aws, docker, ci/cd, observability, monitoring
why not just use cursor/bolt/v0?
you can. many founders do. but here's what happens next:
—your first paying customer sees gibberish instead of a response. the llm hallucinated and there's no retry logic, no fallback — just a broken experience.
—you demo to investors and it works perfectly. launch day hits, 50 users show up, and the agent crashes because the generated code has no rate limit handling.
—you need to add one feature and realize you can't. ai tools generate code — they don't explain the architecture or help you extend it. you're stuck.
i've fixed all of these. $500 gets you production ai features that actually work when users show up.
how it works
1
15-min call
you explain what you're building. i ask questions. we see if it's a fit.
2
scope and quote
i send you what i'll build, how long it'll take, and any constraints.
3
i build, you review
async updates. you see progress. you can course-correct early.
4
handoff
you get the code, docs, and a walkthrough. you're not stuck.
common questions
what if i need more than 20 hours?
we continue at $35/hour. most projects need 40-80 hours total for a complete feature. the 20-hour kickstart lets you evaluate my work before committing to more.
what timezone are you in?
ist (india). i'm flexible with async communication and can overlap with us/eu hours for calls when needed. most updates happen via slack or github.
how do i know you won't disappear?
you get daily async updates, access to the git repo from day one, and docs as i build. if i vanish, you have everything you need to continue without me.
what's the turnaround time?
20 hours of work typically spans 1-2 weeks depending on complexity and feedback cycles. urgent projects can be faster — let's discuss on the call.
do you work with non-technical founders?
yes. i'll explain technical decisions in plain english and make sure you understand what's being built. you don't need to know how llms work — that's my job.
what if the project scope changes?
happens all the time. we'll re-scope and i'll give you a new estimate. no surprises — you approve any additional hours before i start.
can you work with my existing codebase?
yes. typescript/javascript is my specialty, but i can work with other languages too — just need a bit more time upfront to understand the codebase properly.
why $35/hour?
you're paying for ai expertise, not just code. i contribute to the frameworks i use, debug issues others can't, and ship production-ready features — not prototypes that break. i'm building my consulting practice, so you get senior-level ai work while i build my client base.
do you offer project-based pricing?
yes. if you have a well-defined scope, i can quote a fixed price for the whole project. let's discuss what works best for you on the call.
what's your availability?
i take on 1-2 projects at a time to ensure quality. if i'm fully booked, i'll let you know upfront and can schedule for when i'm free.
ready to add ai to your product?
15 minutes is enough to know if we're a fit. no sales pitch — just a conversation about what you're building and whether i can help.
worst case: you lose 15 minutes. best case: your ai feature ships next month instead of next quarter. and if you're not satisfied with my work — full refund, no questions asked.
agencies: i also white-label for client projects. let's talk