ALPHALAB: Autonomous Multi-Agent Research Across Optimization Domains with Frontier LLMs — Brendan Rappazzo
AI Engineer World's Fair 2026 · 20:07
← All organizationswww.morganstanley.com ↗
Financial Services
Morgan Stanley
Morgan Stanley provides investment banking, securities trading, wealth management and investment management for individuals, corporations, governments and institutions. It advises on mergers, acquisitions and capital raising, helps clients manage investments, and provides institutional access to financial markets. Morgan Stanley at Work combines financial advice and technology for employers and employees, while Morgan Stanley Electronic Trading offers algorithms, smart order routing and direct market access across cash equities, options and futures.
Led by Chairman and Chief Executive Officer Ted Pick, the firm also develops technology for financial applications. It originated Morphir and contributed it to FINOS. The open-source framework represents business logic as data, allowing teams to model, share and execute domain knowledge across languages and platforms. This approach makes financial rules reusable across technology systems and gives business and engineering teams a shared representation of those rules.
The company reported $10 trillion in total client assets across Wealth and Investment Management at the end of Q2 2026. Its completed acquisition of EquityZen in 2026 added a private-company share platform to its private-market services, connecting clients seeking liquidity with investors seeking exposure to private companies.
2 talks
Newest first
2 speakers at AIE
Affiliations reflect their AIE appearances, not necessarily current employment.
Start here
- What RL Means for Agents
Start here for a conceptual account of how reinforcement learning could change the relationship between engineered workflows and agent autonomy.
Will BrownAI Engineer Summit 2025
- ALPHALAB: Autonomous Multi-Agent Research Across Optimization Domains with Frontier LLMs — Brendan Rappazzo
Use this talk to understand how a provider-agnostic research harness coordinates worker agents through Kanban-style workflows and Slurm-managed GPU clusters.
Brendan RappazzoAI Engineer World's Fair 2026
Messages from the stage
Evaluation foundations for autonomous experiments
Rappazzo’s AlphaLab presentation emphasizes domain research and trustworthy evaluation construction as the system progresses toward larger-scale worker-agent experimentation. Evaluation failures are part of the account of improving autonomous research.
Learning through reasoning and rewards
Brown discusses reasoning models, test-time scaling, DeepSeek-R1 and GRPO, alongside experimental reward shaping. His talk considers end-to-end reinforcement learning for tool-using agents through the example of OpenAI deep research.
Affiliations reflect each recorded session, not necessarily current employment.