README

Summary

Sachin Bhat — AI/quant engineer; I like building fast, reliable tooling (pricing, agentic routers, TUIs) with operational guardrails.
Experience across agent frameworks, orchestration layers, and production assistants with measurable latency/cost wins.
Background: B.Sc. Data Science & AI (NTU); roles at Tamira, Schaeffler, Dell.

Tamira (AI & Data Engineer, Python/Rust): Rust agentic router (≈3× latency reduction), declarative recipe framework (40–60% token savings), role-aware TUIs with guardrails, automated recipe vulnerability scanner.
Schaeffler (Innovation Engineer): bin-picking/pose-estimation (FoundationPose/BuildSDF), multimodal agents (Llama/Mistral) for task planning, “Iteration of Thought” architecture for behavior trees/code synthesis/video imitation.
Dell (Gen-AI Scientist): async FastAPI orchestration, custom RAG fusion (hallucination ~0.3%), in-house IR replacing paid service (> $1.5M/yr savings), multiple patent-pending frameworks.

Python for quant/prototyping; Panel/Bokeh for quick interactive dashboards.
Rust/Go for services when I need speed and safety.
Data/infra: Postgres, Redis, httpx, uv; IaC and automation where it saves time.
Frontends: fast static sites, minimal JS, purposeful theming (like this neovim vibe).
Topics: option pricing, scenario/risk tooling, automation, reproducible research.

Trade Pricer Platform: EUR option pricer with scenario shocks, heatmaps, Panel UI; CLI + Python API (BSD-3-Clause).
Agentic/router work: Rust-based agent router + orchestration; MCP terminal assistant with guarded shell tools.
Research tooling: RAG chunking explorations (Tattva, Unarxiver), Neural Memory Retriever for long-context reasoning.
Open source: window manager/tiling contributions (labwc, more coming soon...), utilities (cursor-acp, langchain-nmret, stela-mcp), plus assorted forks (bat/coreutils/paru/lemurs).