Search Talent

Run your first talent search in DINQ: describe who you are looking for or upload a JD, answer guided follow-up questions, confirm the search prompt and start searching, review candidate results, then save the best matches to a shortlist.

Step 1: Describe Who You Want to Find

Go to Search and describe your hiring need in natural language.

You do not need a polished job description. Write the way you would brief a teammate.

Example:

I’m looking for an engineer who has worked on LLM inference optimization, ideally with CUDA, TensorRT, or vLLM experience. Open-source work is a plus. Location is flexible, with a preference for North America or remote.

You can also start simple:

Find me 20 engineers working on AI agent infrastructure, ideally with open-source contributions.

The more context you give, the better DINQ can search. Useful details include:

• Role or direction: AI researcher, ML engineer, infra engineer, quant researcher.
• Key skills: LLMs, CUDA, React, TypeScript, PyTorch, robotics.
• Experience level: senior, founding engineer, team lead, startup background.
• Location: Bay Area, New York, London, Singapore, remote.
• Hidden criteria: built from zero to one, strong open-source footprint, highly cited research.

If you already have a JD, you can give it directly to DINQ:

• Paste a text JD.
• Upload a PDF JD.
• Upload an image or screenshot JD.

DINQ reads the JD first, then turns it into a search-ready talent description.

Step 2: Choose Whether to Refine the Search

After you enter a search prompt or upload a JD, the DINQ Agent checks whether the current information is enough to start searching.

If the information is enough, you can start the search right away. If DINQ thinks important details are still unclear, it asks whether you want to keep refining the search. You can continue with guided questions or skip refinement and start searching directly.

When you choose to keep refining, DINQ guides you step by step with AI follow-up questions. It asks one question at a time and provides a few options. You can choose an option or enter your own answer if none of the options fit.

DINQ may ask:

• What role or direction are you hiring for?
• What key skills are required?
• Do you care more about open source, papers, startup experience, or big tech experience?
• Where should candidates be located?
• How many candidates do you want?
• Are there any must-have or must-exclude conditions?

Example:

DINQ: Which background matters most?
A. Open-source contributions
B. Top conference papers
C. Big tech experience
D. Startup zero-to-one experience
E. I want to type my own answer

This flow is useful for first-time users or when your need is still fuzzy. You do not need to know how to write a complete prompt at the beginning. DINQ helps decide whether more detail is needed and breaks missing conditions into step-by-step questions.

Step 3: Confirm and Start Searching

Before the search starts, DINQ summarizes the current search need into a complete search prompt.

You can preview it before the search starts.

If something is wrong, edit it or keep refining. Once you confirm, DINQ starts the search.

After the search starts, DINQ automatically creates a search plan from the final prompt and shows the search process and execution progress in detail.

The search may take some time. DINQ shows the search plan and execution process so you can see how it is looking for candidates, which sources it is checking, and why it changes direction.

DINQ does not only return keyword matches. It tries to connect scattered public signals such as professional profiles, GitHub, papers, personal sites, and public projects.

Step 4: Review Candidate Results

When the search finishes, you will see a set of candidates.

Candidate results may include:

• Name.
• Current role and company.
• Location.
• A short note about why the person may be relevant.
• Public profile links.
• A profile you can open for deeper review.

If someone looks relevant, open the profile and review the details.

Step 5: Review Candidate Enrich Information

Candidate enrich information helps you answer one question: is this person worth a closer look, and are they worth contacting?

Focus on:

• What they have built.
• Whether they have real GitHub contributions.
• Whether their papers or research direction match your need.
• Whether their public history supports the role you are hiring for.
• Whether contact information is available.
• What context you could use for cold outreach.

Do not stop at titles. DINQ is most useful when it helps you see real work beyond the resume.

Step 6: Save Candidates

When someone looks promising, save them to a shortlist.

Shortlists are useful for:

• A role-specific talent pool.
• Candidates from one search.
• People to share with a hiring manager.
• Profiles that need more verification or outreach.

Step 7: Refine the Search

If results are not quite right, add more instructions.

Examples:

Find more people with open-source projects.

Don’t focus only on big tech backgrounds. Find people from startups who have built similar systems.

Expand the location to Europe and remote.

DINQ works well when you refine the search through conversation.