Spaces:
Running
Running
Commit History
-
Upload folder using huggingface_hub
Verified signatureb46865b -
Upload folder using huggingface_hub
Verified signature23b7adb
-
Upload folder using huggingface_hub
Verified signaturef47b400 -
Upload folder using huggingface_hub
Verified signature2e206ee -
Upload folder using huggingface_hub
Verified signature500ff2d -
Upload folder using huggingface_hub
Verified signature24738ba
-
Upload folder using huggingface_hub
Verified signature25bae15 -
Upload folder using huggingface_hub
Verified signature1a247f1 -
Upload folder using huggingface_hub
Verified signature8a9c497 -
Upload folder using huggingface_hub
Verified signature4fe5421 -
Upload folder using huggingface_hub
Verified signature11d0f67 -
Upload folder using huggingface_hub
Verified signature2ca978c -
Upload folder using huggingface_hub
Verified signature41f4b01 -
Upload folder using huggingface_hub
Verified signature79dfd69
-
Upload folder using huggingface_hub
Verified signaturea2b0b5e -
Upload folder using huggingface_hub
Verified signature8d2ff70 -
Upload folder using huggingface_hub
Verified signature1ff2311 -
Upload folder using huggingface_hub
Verified signaturee13b38a -
Upload folder using huggingface_hub
Verified signature1533d3f
-
Upload folder using huggingface_hub
Verified signatureafd0cbd -
feat: improve MLflow model logging to train, evaluate, and compare multiple models, log dataset details, and promote the best version
Saibalaji Namburi4468cb1 -
feat: integrate MLflow model registry, DagsHub metrics tracking, and local Grafana observability stack
Saibalaji Namburi85bb85c -
fix: simplify session bar — remove token display/copy/refresh, auto-width dropdowns, clean auth badge
Saibalaji Namburib19b35f
-
fix: replace raw JWT key display with green/red status badge, add Copy Token btn, fix Redpanda status label
Saibalaji Namburi1a90339 -
feat: initialize DVC, save churn dataset CSV, and enhance CML reporting with performance charts
Saibalaji Namburi8442c59 -
fix: resolve token generation crash in production space and improve JS error handling
Saibalaji Namburi5e79601 -
fix: align customer tier enums, resolve JWT signature mismatch, and update console layout
Saibalaji Namburi7074f09 -
feat: add premium SPA operations console web interface on root route
Saibalaji Namburi80ee821 -
ci/cd: run CML on pushes, redirect space root to docs, and add conftest setup logs
Saibalaji Namburi2d87196 -
fix: conditionally omit X-Frame-Options DENY in HF Spaces
Saibalaji Namburice1b8da
-
fix(lint): resolve ruff code formatting and unused import warnings
Saibalaji Namburi15c00a9 -
feat(ci): configure CI, Hugging Face deploy, and CML model training workflows
Saibalaji Namburi8c62ecf -
feat(k8s): deploy api to kind multinode cluster and verify with adversarial red-teaming simulator
Saibalaji Namburi5553f71
-
feat(phase12): Supabase persistent store + Row Level Security
Saibalaji NamburiSchema (supabase/migrations/001_initial_schema.sql): - tenants: B2B customer organisations - tickets: persistent triage results (status, priority, category, scores) - constitutional_violations: audit log of every safety rule violation - kb_articles: per-tenant knowledge base for RAG grounding - audit_log: immutable append-only event log (RLS: service role INSERT only) - RLS policies on all tenant-scoped tables (app.tenant_id context var) - Auto-updated_at triggers on mutable tables - Indexes: tenant+status, tenant+created, priority, hitl_required Repository layer (src/db/repository.py): - TicketRepository: create, update_status, get, list_recent, log_violation, write_audit - TenantRepository: get_by_slug, create - RepositoryError: typed exception wrapping Supabase errors - RLS double-check at app layer (eq tenant_id on every query) - Non-fatal audit log writes (never crash triage on audit failure) Tests: 18 new tests, all passingaf62db2 -
feat(observability): upgrade Langfuse tracing per langfuse/skills best practices
Saibalaji NamburiApplied all baseline requirements from github.com/langfuse/skills: - Descriptive trace name: 'ticket-triage' (was generic 'triage') - Explicit trace input: only ticket_text + channel (not all function args) - PII masking: email/phone/SSN/IBAN/card stripped before sending to Langfuse - Singleton client: one Langfuse() per process (not one per request) - flush via client.flush() not trace.flush() (SDK v3 best practice) - agent_span accepts input_data param for meaningful span inputs - Generation names use hyphens: 'gemma3:4b-generation' (was underscore) - Prompt messages and completions PII-masked before logging - Tags use bounded cardinality: [tier, channel] not tenant_id5e56072 -
feat(phase11): Langfuse LLM tracing + Constitutional Policy Engine
Saibalaji NamburiLangfuse Observability (src/monitoring/langfuse_tracer.py): - TriageTrace: trace/span/generation hierarchy per triage request - Graceful no-op degradation when LANGFUSE_PUBLIC_KEY not set - LiteLLM callback registration (auto-traces all LLM calls) - Quality scores: constitutional_compliance, resolution_quality, rag_grounding - Wired into API lifespan startup and triage background task Constitutional Policy Engine (src/responsible_ai/constitutional_policy.py): - 7 constitutional rules: PII_PROTECTION, AI_IDENTITY, TOXICITY_GUARD, NO_COMMITMENTS, SCOPE_LIMITATION, LANGUAGE_CONSISTENCY, COMPETITOR_NEUTRAL - 3 severity levels: CRITICAL (block), VIOLATION (regenerate/redact), WARNING (warn) - Fast regex-based checks (<5ms) for all rules - PolicyVerdict with score (0-1), action, violations, warnings, safe_fallback - Critical violations auto-trigger HITL escalation - Runtime rule addition/removal support Tests: 38 new tests, all passingbf95443 -
feat(phase10): FastAPI REST API + Dockerfile + SSE streaming
Saibalaji Namburi- src/api/main.py: FastAPI app factory, lifespan, CORS, security headers, structlog middleware, rate limiting (slowapi), global exception handler - src/api/auth.py: JWT authentication, tenant_id extraction, RBAC (require_role) - src/api/models.py: Pydantic v2 API request/response schemas (7 models) - src/api/store.py: In-memory triage store (Supabase PostgreSQL drop-in for Phase 12) - src/api/routers/triage.py: POST/GET /triage, HITL resume, tenant isolation - src/api/routers/health.py: Kubernetes liveness (/health) + readiness (/ready) probes - src/api/routers/stream.py: SSE real-time triage progress stream - src/api/routers/metrics.py: Prometheus metrics (counters + histograms) - Dockerfile: Multi-stage build, non-root user, JDK for PySpark, HEALTHCHECK - Dockerfile.hf: Lightweight HF Spaces image (port 7860, no PySpark) - tests/unit/test_phase10_api.py: 20 tests — all passing Routes: POST /triage, GET /triage, GET /triage/{id}, POST /triage/{id}/resume, GET /triage/{id}/s…8ce332b -
feat: initial commit — CustomerCore Phases 1-9 complete
Saibalaji Namburi- Phase 1: Project scaffold, Docker Compose (6 services), Doppler secrets - Phase 2: Redpanda streaming — 4 event producers (tickets, billing, incidents, products) - Phase 3: PySpark Bronze→Silver pipeline, Presidio PII masking, MinIO lakehouse - Phase 4: dbt Gold analytics layer — 7 business marts, 18 data contract tests, DuckDB - Phase 5: Cryptographic Privacy Vault — AES-256-GCM, RBAC, GDPR Article 17 right-to-erasure - Phase 6: Multi-tenant Hybrid RAG — ChromaDB + BM25 + RRF fusion + 3-layer semantic cache - Phase 7: SLA-aware multi-model LLM router — LiteLLM, OpenRouter, 54% local routing - Phase 8: Multi-language support (7 langs, 87k rows) + Unified B2B Graph-RAG (NetworkX) - Phase 9: LangGraph 6-agent supervisor — HITL, MemorySaver checkpointing, 223 tests Stack: Python 3.12, Redpanda, MinIO, ChromaDB, Redis, DuckDB, dbt, LangGraph, LiteLLM Dataset: 52,417 real-world support tickets (Bitext EN + Amazon MASSIVE DE/FR/ES/PT/IT/NL/JA)3cfefe7