Agentic AI · MLOps · Quant Systems · Memory-layer R&D
I design AI systems as composable architectures: agents with explicit state, durable memory, evaluation loops, observability and deployment boundaries. The work here is organized around how systems are structured, not just what they output.
flowchart LR
Goal["Research or product goal"] --> Workflow["Agent workflow"]
Workflow --> Memory["Memory layer"]
Workflow --> Tools["Tools and data"]
Workflow --> Eval["Evaluation gate"]
Eval --> Observe["Traces and monitoring"]
Observe --> Learn["Iteration loop"]
Learn --> Workflow
Most agent projects fail at the seams: memory leaks into prompts, evaluation happens after release, and deployment is treated as an afterthought. I treat those boundaries as first-class design problems.
- Agentic orchestration — explicit state, bounded tool use, human oversight and traceable decisions.
- Memory design — episodic, semantic and graph memory with conflict resolution instead of infinite context growth.
- Evaluation-first ML — metrics, gates and regression checks before a system is trusted.
- Quant research systems — backtesting, risk constraints and clear separation between research and execution.
- Operational shape — APIs, containers, CI and runbooks that a small team can actually run.
| Repository | Architectural theme |
|---|---|
agentic-quant-lab |
Planner → simulator → risk engine → research report |
memory-layer-rnd |
Multi-store memory with retrieval and conflict resolution |
agentic-mlops-foundry |
API boundary, runtime, eval gate and deployment path |
llm-finetuning-eval-lab |
Data → baseline → metrics → model card → CI gate |
Agents: LangGraph-style workflows, role decomposition, RAG, tool calling, eval harnesses
ML/MLOps: Python, FastAPI, Docker, GitHub Actions, experiment tracking
Quant: pandas, NumPy, backtesting, risk metrics, RL research
Systems: API design, ADRs, test strategy, operational runbooks
- Architecture should make failure modes visible early.
- Memory is a contract, not a transcript dump.
- Agent autonomy requires evaluation, traces and rollback paths.
- Finance systems must separate research, paper trading and live execution.
- Good engineering shows up in docs, tests and explicit trade-offs.