Work

A live Kalshi volatility stack built around streaming BTC data, feature engineering, and model-driven execution with real capital on the line.

Core challenge: Pricing short-horizon volatility under noisy live conditions instead of relying on retrospective backtests alone.

• Kraken WebSocket ingestion
• Kalshi order-book signal design
• LightGBM plus PatchTST execution loop

A Rust-first execution system for mirroring high-signal wallets on Polymarket with low latency, deterministic handling, and operator-visible control loops.

Core challenge: Translating on-chain observation into mirrored positioning fast enough to matter without making the system operationally opaque.

• Rust-first execution engine
• Signal-aware wallet mirroring
• Latency-aware trade handling

An inference runtime for constrained hardware that combines speculative decoding, smart routing, and KV-cache compression to make smaller devices more useful.

Core challenge: Pushing more reasoning capability through limited hardware without pretending compute budgets do not exist.

• Speculative decoding path
• KV-cache compression layer
• Runtime-level routing logic

A multi-agent orchestration layer that distributes work across Codex, Claude, Gemini, and Kimi so parallel model effort becomes a usable execution workflow.

Core challenge: Making heterogeneous model strengths additive, structured, and operational instead of noisy.

• Parallel agent dispatch
• Cross-model synthesis
• Dependency-aware task routing

A multi-agent coding roundtable that turns model disagreement into a visible review process for architecture, implementation planning, and technical critique.

Core challenge: Preserving shared context while letting specialist agents critique the same problem from meaningfully different angles.

• Shared context handoff
• Multi-agent review ritual
• Architecture-oriented critique loop