marko-k · HackerTrans

I agree with everything you've said here. I've scanned it many times myself, and it's obviously brilliant. I probably just chose my words poorly.

What I meant was that "Stopping by Woods" is so widely read that I wanted to share a few other poems that don't get as much attention.

marko-k

昨年議論

He's a brilliant poet, though often misunderstood. Skip Stopping by Woods on a Snowy Evening and try Birches [1], A Tuft of Flowers [2], or the The Witch of Coös [3] instead.

[1] https://www.poetryfoundation.org/poems/44260/birches

[2] https://www.poetryfoundation.org/poems/44275/the-tuft-of-flo...

[3] https://www.poetryfoundation.org/poetrymagazine/browse?volum...

marko-k

昨年議論

If Responses is replacing Assistants, is there a quickstart template available—similar to the one you had for Assistants?

https://github.com/openai/openai-assistants-quickstart

marko-k

昨年議論

  Location: New York City, NY. 
  Remote: Yes  
  Willing to Relocate: No  
  Technologies: LLM & AI APIs (OpenAI, Anthropic, Structured Outputs, Tool/Function Calling, Assistants), ReactJS, NextJS, Python, Django, NodeJS, SQL, JavaScript/TypeScript, Stripe  
  Résumé/CV: Available upon request  
  Email: [email protected]

I've built viral products and lead development on AI-powered tools. I'm a senior software developer specializing in AI applications for education and document processing. I code across the entire stack, have sharp design intuition, and deliver standout UI/UX. I also regularly run technical workshops and communicate complex ideas clearly.

marko-k

昨年議論

I’d pivot away from targeting developers and focus on "normal" power users—maybe even pick a specific vertical. Most devs won’t pay for a form builder, but people *will* pay for beauty and ease of use (which is why Typeform works). If you want paying users, you’ll probably need to add integrations so they don’t have to mess with the backend.

Also, I’d rethink the UX for the AI era. Instead of a traditional form builder, what if users just *described* the form they need, and AI generated it instantly? Then they could refine it with natural language—“Add an email field” or “Make question 3 optional”. That way, they go from zero to a polished form in seconds.

You already have the structured data and the code logic—you'd just have to build the AI integration. Luckily, most of the models can speak JSON now with structured outputs, so it's entirely possible.

marko-k

昨年議論

  Location: New York City, NY. 
  Remote: Yes  
  Willing to Relocate: No  
  Technologies: LLM & AI APIs (OpenAI, Anthropic, Structured Outputs, Tool/Function Calling, Assistants), ReactJS, NextJS, Python, Django, NodeJS, SQL, JavaScript/TypeScript, Stripe  
  Résumé/CV: Available upon request  
  Email: [email protected]

I've built viral products and led development on AI-powered tools. Senior software developer with extensive experience in AI tooling, particularly structured outputs and applications in education, text, and document processing. I can code across the stack and have a keen sense for design, UI, and UX. I also run technical workshops for large groups and communicate complex topics clearly.

marko-k

昨年議論

If that were the case, they’d be neck and neck with Anthropic and Claude. But ChatGPT has far more market share and name recognition, especially among normies. Branding clearly plays a huge role.

marko-k

2 年前議論

Not OP, but I work on AI in higher ed at a major university.

I get the concerns about AI grading. The solution isn't to have AI grade entire assignments at once. Instead, break down the assessment into smaller, discrete tasks and develop a grading rubric around those. The idea is to limit how the AI can respond - usually to simple binary choices like completed/not completed, true/false, etc. (Also, the models have been RLHF’d to generally put a positive spin on things, so if anything they’re likely to be overly generous in assessment.)

From there, provide the AI with the answer key, student response, rubric, and any other necessary context then use the Structured Outputs API to force consistent responses for each discrete task. I've had the most success using boolean values or simple enums (like "Correct", "Partially Correct", "Incorrect"). You can include a field for reasoning, then chain AI calls to get a second assessment as verification.

That's the high-level gist of it, though I'm skipping a lot of details. I have a basic demo of how this works on my site if you're interested: https://www.markokrkeljas.com/projects/real-time-task-tracki...

コメント