Session Canvas
5 proposals are waiting for a decision.
Showing 25 of 25 submissions. Click any row to open the full proposal and record a decision.
| Vector Databases: Choosing One You Won't RegretRohan Mehta | Accepted queue |
| Guardrails at the Edge: Safety Without the Latency TaxMarcus Lindgren | Pending |
| A Practical Taxonomy of LLM Failure ModesNadia Haddad | Declined |
| Cutting Inference Costs 70% Without Touching the ModelJonas Weber +1 | Accepted |
| Prompt Caching, Batching and Other Boring WinsKwame Boateng | Accepted |
| Shipping Agents That Don't Melt Down at 3 A.M.Samuel Adeyemi | Accepted |
| Why Your Chatbot Needs a Product ManagerAiden O'Rourke | Declined |
| The RAG Is Dead, Long Live RetrievalAmara Okonkwo | Accepted |
| Structured Output Is a Distributed Systems ProblemMei-Lin Chao | Accepted |
| Human-in-the-Loop Without Being Human-HostileGrace Oyelaran +2 | Accepted |
| Multi-Agent Orchestration: A Field ReportAiden O'Rourke | Accepted queue |
| Fine-Tuning Is a Data Problem, Not a Model ProblemInes Duarte | Accepted queue |
| The Migration We Regret: Rewriting Our Inference StackJonas Weber | Withdrawn |
| Building a Tool-Calling Runtime in 300 LinesTobias Andersson | Accepted |
| Streaming UIs for Agentic WorkflowsGrace Oyelaran | Pending |
| The Economics of Context WindowsJonas Weber | Accepted |
| Secrets, Sandboxes and Untrusted Tool CallsMarcus Lindgren | Pending |
| Teaching Your Codebase to Answer QuestionsRohan Mehta | Pending |
| Observability for Nondeterministic SystemsYuki Tanaka | Accepted |
| Evals as a Product Surface: Making Quality LegibleDiego Ferreira | Accepted |
| From Notebook to Nine-Nines: Productionising MLElena Vasquez | Accepted |
| Running Open Models on Commodity GPUsKwame Boateng | Draft |
| Benchmarks We Actually BelieveDiego Ferreira | Draft |
| Semantic Diffing for Code ReviewRohan Mehta | Draft |