Kalletlamadhav/sql_optimized_env_new

1---2title: SQL Optimization RL Environment3emoji: 🗄️4colorFrom: blue5colorTo: green6sdk: docker7app_port: 78608tags:9  - openenv10  - sql11  - rl12  - optimization13---14 15# 🗄️ SQL Query Optimization RL Environment16 17An **OpenEnv-compliant Reinforcement Learning environment** for training agents to automatically optimize SQL queries on **real-world Indian government-scale datasets**.18 19---20 21## 🚀 Why This Project?22 23Modern databases often suffer from:24- Slow queries due to poor indexing25- Inefficient joins and anti-patterns26- High latency on large-scale datasets27 28This environment enables:29👉 Training AI agents to **detect and fix SQL inefficiencies automatically**  30👉 Benchmarking query optimization strategies in a **realistic setting**31 32---33 34## 🏛️ Data Domains (Realistic Scale)35 36| Domain  | Tables | Description |37|---------|--------|-------------|38| **GST** | `gst_invoice_records`, `gst_invoice_items` | B2B invoices, HSN items, tax splits |39| **PDS** | `ration_card_beneficiaries`, `pds_allotments` | Ration distribution system |40| **Railway** | `railway_trains`, `railway_pnr_bookings` | Ticket booking workloads |41| **MGNREGA** | `mgnrega_workers`, `mgnrega_attendance`, `mgnrega_payments` | Rural employment data |42 43---44 45## 🎯 Task Design (9 Tasks × 5 Levels)46 47| Level | Tasks | Anti-Patterns |48|-------|-------|---------------|49| 1 Intro | `pds_select_star` | SELECT_STAR |50| 2 Easy | `gst_missing_index`, `railway_simple_filter` | MISSING_INDEX |51| 3 Medium | `gst_n_plus_one`, `pds_cartesian`, `mgnrega_wildcard` | N+1, CARTESIAN, WILDCARD |52| 4 Hard | `gst_multi_join`, `railway_tatkal_workload` | Multi-pattern |53| 5 Expert | `mgnrega_schema_e` | UNBOUNDED_AGG + CAST |54 55---56 57## ⚙️ System Architecture58 59```text60Agent (LLM / RL Policy)61        ↓62   OpenEnv API63        ↓64SQL Execution Engine (SQLite)65        ↓66Execution Plan + Metrics67        ↓68Reward Function (5 factors)