SHIVASHANKAR S MENON / WORK
Work
Systems that can be measured — retrieval pipelines, multi-agent engines, and the evaluation harnesses that keep them honest.
3 PROJECTS · 2023 — ONGOING · 2 PEER-REVIEWED
| P-01 | Amphoreus D&D — Multi-Agent LLM Simulation | AI ENGINEERING | ● ONGOING · 2026 |
| P-02 | Accessible Wheelchair Routing | ML RESEARCH | PUBLISHED · 2024 |
| P-03 | Foreground Emissions Modeling for Astronomy | ASTRONOMY / ML | PUBLISHED · 2023 |
Amphoreus D&D — Multi-Agent LLM Simulation
2026 · ONGOING · PYTHON / CHROMADB / MULTI-PROVIDER LLM
plate 1 — engine demo · click to play
| ORCHESTRATION | Three distinct cognitive roles coordinated by a 4-stroke turn loop; world state mutates only through typed, logged tool calls, never parsed prose. |
| MEMORY | Agents pull top-K relevant context from owned, vector-indexed stores (ChromaDB); provider-independent, so any role can run a different model. |
| EVALUATION | A variance-aware harness (N≥3 per condition, distributions not point values, an independent LLM-judge, pre-registered thresholds) to help decide what to build by measuring first. |
Accessible Wheelchair Routing
2024 · PUBLISHED · PYTORCH / TRANSFER LEARNING / SENSOR DATA
Built two datasets of vibration data from manual and powered wheelchairs across fifteen surface types. Trained a surface-classification model using adaptive activation functions and extended it to powered wheelchairs via transfer learning, reaching 98.8% accuracy — outperforming SOTA by 3% with 40% less train time.
Foreground Emissions Modeling for Astronomy
2023 · PUBLISHED · ROBUST REGRESSION / GALEX TELEMETRY
Parsed and analyzed deep-sky observation data from the GALEX satellite's spacecraft state files. Derived an empirical model of UV foreground emissions using robust regression, and generalized it to deep and medium-sky observations.