Infinity Reach - The Digital Marketing Agency
Website:
infinityreach.com
Job details:
We are not hiring a coder. We are hiring someone who builds like they have a team of ten in their laptop.
If you have ever stayed up until 4am because the build finally worked, if you have forty half-finished repos and three that are actually live, if you run models locally because you wanted to see what would happen, if you read the changelog for fun: this is for you.
We don't care where you studied, whether you graduated, or how many years you have on a resume. We care about what you have built, how fast you built it, and how good it actually is.
What this job actually is
We build products. Lots of them. Web, mobile, agents, internal tools, weird experiments that turn into real things. Some of them will be big. Some will die in a week. All of them get built fast and shipped.
Your job is to take an idea and turn it into something live, using every AI tool available to you, and then make it good enough that it holds up under real users. You will architect it, orchestrate the agents that write most of it, review what they produce, harden it, and ship it.
Typing speed is irrelevant. Judgment is everything.
What you will do
Ship things. Idea to live product, repeatedly. Web apps, mobile apps, tools, agents, whatever the problem needs. You own it end to end.
Direct AI, don't babysit it. Run agentic coding setups seriously. Multiple agents, custom system prompts, MCP servers, skills, subagents, background jobs. Build the workflow, then keep sharpening it.
Engineer context. Structure codebases and data so models stay grounded instead of drifting into confident nonsense at hour three. This is a real skill and it is one of the main things we are hiring for.
Design the data. Schemas, relationships, indexes, pipelines. Turn messy real-world input into clean structured data that stays fast and cheap as it grows.
Run models locally. Deploy, benchmark, and integrate open-weights models for privacy, latency, and cost. Know when a local 7B is enough and when you need to call out to a frontier model.
Be the adult in the room. Review and refactor AI-generated code. Catch the security hole, the N+1, the thing that works today and collapses at 10,000 rows. Ship code you would be happy to inherit.
Pick up whatever is new. New model drops, new framework, new paradigm. You should be the person who has already tried it by the time the rest of us hear about it.
The stack, today
Here is what is on the bench right now. A chunk of it will be different in six months. That is the point. We are not hiring for a stack, we are hiring for someone who eats new stacks for breakfast.
Web: Next.js, React, TypeScript, Tailwind, shadcn/ui, Astro, Svelte, and whatever ships fastest.
Mobile: React Native, Expo, Flutter, native when it matters
Backend and data: Firebase, Firestore, Postgres, Supabase, Drizzle, edge and serverless functions, Cloudflare Workers.
AI layer: Anthropic, OpenAI and Gemini APIs, structured outputs, tool use, RAG, embeddings, vector search (pgvector, Pinecone, Turbopuffer, Weaviate), evals and prompt testing
Agentic dev tooling: Claude Code, Cursor, Windsurf, Antigravity, MCP servers, custom skills and system prompts, multi-agent orchestration
Local models: Ollama, LM Studio, vLLM, llama.cpp, DeepSeek, Qwen, Llama, Mistral, quantization, LoRA and fine-tuning if you are into it
Ingestion and automation: Playwright, Puppeteer, scrapers, feeds, ETL, LLM-based extraction and entity resolution
Infra: Vercel, Cloudflare, GCP, Docker, GitHub Actions, CI/CD
Whatever is next: voice, realtime, computer use, browser agents, on-device inference, WebGPU, and the thing that gets announced next month that we haven't heard of yet
You are not expected to know all of this. You are expected to be the kind of person who could learn any of it in a weekend.
Who we are actually looking for
You probably recognize yourself in a few of these:
- You have shipped something real and public, alone, and people used it
- Your GitHub or your portfolio is more interesting than your resume
- You have a homelab, or a Pi doing something stupid, or a server rack in your bedroom that your family has stopped asking about
- You have run a local model just to see how far you could push it
- You have written a custom system prompt longer than most people's cover letters
- You build tools for yourself before anyone asks you to
- You have opinions about tooling and you can defend them
- You get bored when the work is not hard
We are hiring for slope, not intercept. If you are early in your career but obviously dangerous, you are exactly who we want.
Who this is not for
- People who want tickets assigned to them
- People who think AI coding tools are a toy or a threat
- People who need a spec before they can start
- People who measure themselves in years of experience instead of things shipped
- People who want to do the same thing next year that they are doing this year
No shade. This is just a very specific kind of job and it will make you miserable if this is you.
Bonus points
- You have run open-weights models locally for offline or low-latency setups
- You have built an agent, a pipeline, or a tool that does real work unattended
- You have shipped a full app single-handedly with AI acceleration
- You have built something genuinely weird that we will want to ask you about
- You contribute to open source, or you have written about how you build
How to apply
Resumes are optional. Proof is not.
Send us:
- Links to things you have shipped. Live products, apps, tools, experiments. Anything real.
- Your toolchain, in your own words. A short write-up or a 2 minute video or Loom walking through how you actually build: which agents, how you manage context, how you go from idea to shipped.
- The most unhinged thing you have built. Doesn't have to be useful. Just has to be yours.
If you built something at 2am that you are still proud of, lead with that.
Compensation: [If you are looking to apply to this job for money then please don't]
Apply to: [vipul@infinityreach.com]
Click on Apply to know more.