Hayvin's knowledge graph powers different workflows depending on who you are and what you need. Same graph. Same data. Different queries.
Hayvin scores candidates on actual contributions — commit depth, PR quality, paper relevance — not keywords on a resume. Filter by tech stack, recency, location, and availability. No cold InMails.
Query
"Find inference engineers who've contributed to vLLM or TensorRT-LLM in the last 90 days, currently based in SF."
7 matches found
Infra Eng · xAI
MLSys · Anyscale
Research Eng · Together AI
Upload the attendee list or connect your conference badge. Hayvin cross-references attendees against its knowledge graph and surfaces the 10–15 people most likely to have a genuinely valuable conversation with you — before the conference even starts.
Query
"Who at NeurIPS 2025 is working on KV cache optimization or speculative decoding, and open to grabbing coffee in the next 48 hours?"
12 matches found
Research · DeepMind
Eng · Mistral AI
PhD · CMU
The knowledge graph maps citation relationships, co-authorship history, and topic clusters — so you can find researchers with complementary expertise, not just overlapping keywords. Identify collaboration gaps and match on methodology as much as subject matter.
Query
"Find researchers working on sparse attention mechanisms who have at least 2 papers on OpenReview and are within 50 miles."
5 matches found
PhD · Stanford
PostDoc · UCB
Research Sci · Google Brain
Sometimes the most valuable conversation is with someone who just solved the same problem you're stuck on. Hayvin surfaces engineers based on what they're actively building — not their job title or employer.
Query
"Who's deploying LoRA fine-tuning in production and has GitHub activity on PEFT or Axolotl in the last 60 days?"
9 matches found
MLOps Eng · Cohere
AI Eng · Replicate
Indie Researcher
