Retrieval-Augmented Generation (RAG) × Developer Tools

Sakana AI opens up its product team

Sakana AI opens up its product team

✎ Story body

Turning research into a product is a separate job at Sakana AI, and the lab has now published who does it and how they are hired.

What happened

Sakana AI published an FAQ-style look inside the product team that carries its research into shipped software. It describes two roles: the Applied Research Engineer, who turns a result into something that runs, and the Software Engineer, who builds that into a product. For each it walks through a typical day, how the team is organised, how hiring proceeds, and what the company looks for. The account is drawn from interviews with current members, and it notes that the team works without a block of hours when everyone is required to be present.

Why it matters

The striking part is the split itself. Making a model better and making it usable are treated as two jobs with two titles and two different days, rather than as one role a researcher gradually grows into. Read as an org chart, it suggests that whether a lab ships depends less on model quality than on who carries the translation, and how. What the piece does not show is the consequence: it describes the structure that exists now, not what that structure has produced.

What to watch

Which of the two roles the open positions actually fill first. The count that settles this is not papers, but products that reach users.

▲ Official & Press
Official

Inside Sakana AI's Product Team

Sakana AI Blog (ja) ・ 2026-09-16 ・ 📌

Sakana AI opens up its Product Team: culture, workflow, and hiring FAQ

Official

TensorRT Edge-LLM Completes the MLPerf Edge Agentic Benchmark 6.4x Faster on Jetson AGX Thor

NVIDIA Developer Blog ・ 2026-09-16

NVIDIA runs MLPerf Edge Agentic 6.4x faster on one Jetson AGX Thor

Academic (arxiv etc.) 18 ▾
Academic

When Should LLMs Abstain? Chain-of-Self-Questioning for Selective Risk Control

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-15

Academic

JustFit: 200K-Token LLM Serving on a 24 GiB Laptop with Just-in-Time State Management

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-15

Academic

Tables Decoded: DELTA for Structure, TARQA for Understanding

arXiv cs.LG (Machine Learning) ・ 2026-09-15

Academic

Coding Agents Have Converged: Why the SWE-bench Leaderboard Can No Longer Order Its Top Entries, and What to Measure Instead

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-15

Academic

Where Should a Document Live: Context, Representations, or Parameters?

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-15

Academic

Mo' Models, Mo' Problems: How to best select model pools when designing Multi-Agent Systems

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-15

Academic

Easy to Catch a Liar, Hard to Clear an Honest One: Language Models Diagnosing a Corrupted Reward Channel from a Verified Record

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-15

Academic

Grounding SWE-Agent Decisions in Architecture-0 Design: Navigating Unknown Unknowns through Physical Mapping

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-15

Academic

Shared-Prefix KV Reuse Across Standard LoRA Adapters: Quality and Serving Tradeoffs

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-15

Academic

Symbolic Separation: Grounding Deep Agents in Knowledge Graphs for Trustworthy Operational Data Analytics

arXiv cs.AI (Artificial Intelligence) ・ 2026-09-15

Academic

EviScope: Paired Counterfactual Evidence Diagnostics for Faithful and Efficient Grounded Language Models

arXiv cs.CL (Computation and Language) ・ 2026-09-15

Academic

Diagnosing the Fact-Grounding Gap in Multi-Hop Question Answering

arXiv cs.CL (Computation and Language) ・ 2026-09-15

Academic

The Role of Implicit and Explicit Demographic Signals in Large Language Model-based Student Assessment

arXiv cs.CL (Computation and Language) ・ 2026-09-15

Academic

When Confidence Signals Disagree: Local and Global Confidence in Autoregressive Language Models

arXiv cs.LG (Machine Learning) ・ 2026-09-15

Academic

Lit3R: Retrieve-Relate-Read for Evidence-Grounded Question Answering over Scientific Literature

arXiv cs.CL (Computation and Language) ・ 2026-09-15

Academic

ImpossibleRubrics: Stress-Testing Generated Rubrics as Reward Signals

arXiv cs.CL (Computation and Language) ・ 2026-09-15

Academic

Smarter by the Moment: Environment-Driven Dynamic Policies for Continual LLM Improvement

arXiv cs.CL (Computation and Language) ・ 2026-09-15

Academic

LSREP: A Longitudinal State-Replay Protocol for Evaluating Conversational Memory, with ICE v2 as an Audited Local-First Architecture

arXiv cs.CL (Computation and Language) ・ 2026-09-15

← Story Archive